Reproduction in the U.S., 1965 9781400886517

Table of contents :
Contents
Chapter I. Background of the 1965 National Fertility Study
Chapter II. Orientations Toward Numbers of Children
Chapter III. The Trend of Expected Parity
Chapter IV. Racial, Religious, and Socioeconomic Differences in Fertility
Chapter V. Contraception
Chapter VI. Use of the Pill
Chapter VII. Some Effects of Pill Use
Chapter VIII. Catholic Conformity Before and After the Papal Encyclical
Chapter IX. Fertility Planning
Chapter X. Attitudes Toward Abortion
Chapter XI. Birth Intervals and Pregnancy Intervals
Appendix A: Inconsistency of Response
Appendix B: 1965 Questionnaire
Appendix C: Related Publications
Index

Citation preview

REPRODUCTION IN THE UNITED STATES 1965

PUBLISHED FOR THE OFFICE OF POPULATION RESEARCH PRINCETON UNIVERSITY

REPRODUCTION IN THE UNITED STATES 1965 NORMAN B. RYDER AND CHARLES F. WESTOFF

PRINCETON UNIVERSITY PRESS PRINCETON, NEW JERSEY 1971

Copyright © 1971 by Princeton University Press ALL RIGHTS RESERVED

LCC: 78-120760 ISBN: 0-691-98218-0

Publication of this book has been aided by a grant from the Office of Population Research Princeton University

This book has been composed in linotype Times Roman Printed in the United States of America by Princeton University Press, Princeton, New Jersey

Princeton Legacy Library edition 2017 Paperback ISBN: 978-0-691-62036-7 Hardcover ISBN: 978-0-691-65476-8

Contents I II

Background of the 1965 National Fertility Study

3

Orientations Toward Numbers of Children

19

III

The Trend of Expected Parity

37

IV

Racial, Religious, and Socioeconomic DifiEerences in Fertility 53

V

Contraception

97

VI

UseofthePill

139

Some Effects of Pill Use

155

VII VIII

Catholic Conformity Before and After the Papal Encyclical 185

IX

Fertility Planning

223

X

Attitudes Toward Abortion

267

Birth Intervals and Pregnancy Intervals

295

Appendix A: Inconsistency of Response

359

Appendix B: 1965 Questionnaire

373

Appendix C: Related Publications

415

Index

417

XI

CHAPTER I

Background of the 1965 National Fertility Study THE roots of the 1965 National Fertility Study can be traced at least as far back as 1938, when a committee was formed to launch an inquiry into the social and psychological factors affecting fertility, an effort that eventually produced the five volumes of research reports constituting the now famous Indianapolis Study.1 The original impetus for that study was the growing impression that census statistics on differential fertility were inadequate to provide sufficient understanding of the social and psycho­ logical determinants of reproductive behavior, and that there was a need to collect information beyond the reach of the federal government, on such characteristics as fecundity, contraception, attitudes, and religion. The interest in the subject was much more than academic: the birth rate had been declining for many decades, and there was concern about the possible need for a national population policy to counter the threat of a declining population size. The Indianapolis Study was primarily a hypothesis-testing operation: the original study group formulated twenty-three specific hypotheses about factors affecting the probability of fertility planning and the size of planned families. P. K. Whelpton and Clyde V. Kiser undertook much of the original analysis and continued over the years to collaborate with various other social scientists in the preparation of analytic reports. The direct descendant of the Indianapolis Study was the Princeton Fertility Study,2 so named because the work was done at the Office of Population Research at Princeton University. The Princeton Study in­ corporated many of the hypothesis-testing interests of its intellectual par­ ent, added others, changed the study design from a cross-sectional to a longitudinal format, and broadened the universe for the sample to en­ compass the nation's largest metropolitan areas. Again the interest was 1 Pascal K. Whelpton and Clyde V. Kiser, eds., Social and Psychological Factors Affecting Fertility, 5 vols., (New York: Milbank Memorial Fund, 1946-1958). See also C. V. Kiser and P. K. Whelpton, "Resume of the Indianapolis Study of Social and Psychological Factors Affecting Fertility," Population Studies 7 (1953): 95-110; Ronald Freedman, "American Studies of Factors Affecting Fer­ tility," International Population Conference (New York: International Union for the Scientific Study of Population, 1961). 2 Three volumes report the different phases of the longitudinal study: Charles F. Westoff, Robert G. Potter, Jr., Philip C. Sagi, and Elliot G. Mishler, Family Growth in Metropolitan America (Princeton: Princeton University Press, 1961); C. F. Westoff, R. G. Potter, Jr., and P. C. Sagi, The Third Child (Princeton: Princeton University Press, 1963); Larry Bumpass and C. F. Westoff, The Later Years of Childbearing (Princeton: Princeton University Press, 1970).

CHAPTER J

primarily theoretical: the restriction of the sample to once-married women who had recently given birth to their second child sharply re­ duced the usefulness of the study for generalizations about the reproduc­ tive behavior of American couples in general. Because of such limitations, because there was a need for estimating parameters of demographic behavior that could not be provided by fed­ eral statistical agencies, and because of the interest in a new approach to the problem of forecasting fertility, another offspring of the Indian­ apolis Study was born in 1955. Ronald Freedman and P. K. Whelpton undertook the first Growth of American Families Study, or, as it subse­ quently became known in the trade, GAF-I.3 This was the first attempt to obtain a wide range of demographic and social data on a national scale for the purpose of plotting the future course of fertility. This study was responsible for providing the first estimates for the population of the United States of the incidence of sterility, the proportion of couples us­ ing contraception, the numbers of children wanted and expected, and many other parameters of demographic interest. The "number of chil­ dren expected" measure, devised to provide a basis for estimating com­ pleted fertility and to facilitate the task of population forecasting, is a concept which has now become a standard part of the demographer's lexicon, and is receiving serious attention by professionals in the federal government as a possible aid in the preparation of official forecasts. In 1960, Whelpton, Campbell, and John E. Patterson collaborated in the production of the second Growth of American Families Study, GAF-II.4 This was conceived as a follow-up and extension of the scope of GAF-I; it provided for the first time an opportunity to evaluate the validity of women's fertility expectations and to begin the analysis of time trends in the proportions using contraception, the level of family size desired, and the pattern of group differences in fertility. One impor­ tant innovation of GAF-II was the first inclusion of nonwhite women in the sample, and therefore the first opportunity to estimate parameters for the total population. BEGINNINGS OF THE 1965 STUDY

Since GAF-I and GAF-II had demonstrated their considerable useful­ ness, it may have been taken for granted that the series, begun in 1955 3The study was published as: R. Freedman, P. K. Whelpton, and A. A. Camp­ bell, Family Planning, Sterility and Population Growth (New York: McGraw-Hill, 1959). Ν. B. Ryder was Mr. Whelpton's assistant during the first two years of the 1955 study. 4 P. K. Whelpton, A. A. Campbell, and J. E. Patterson, Fertility and Family Planning in the United States (Princeton: Princeton University Press, 1966).

BACKGROUND

and repeated in 1960, would be continued at regular intervals. However it came to our attention in late 1964 (at a conference on genetics and demography at the Princeton Inn) that nobody had begun to organize the 1965 round. With a strong feeling that this important time series— one of the few in any area of social science with such depth and replica­ tion capability—should not be interrupted, and with the active interest and availability of sources of generous financial support, we decided to undertake the 1965 National Fertility Study (NFS). We learned that the federal government, specifically the National Cancer Institute, would be interested in underwriting a study on the health implications of oral con­ traception, and that they would not be averse to an expansion of the scope of the study to include the demographic interests of GAF-I and GAF-II. Subsequently it developed that the kind of research we were un­ dertaking would not be well-designed to provide the kind of information necessary to resolve medical questions about oral contraception; the aegis of the study then became the National Institute of Child Health and Human Development, and the study design assumed its present form. One of our initial naive assumptions in making this decision was that the style of research, the instrument, and the sample design were sufficiently well-defined and tested that it would be necessary only to re­ charge the battery to make the output flow almost automatically. As we learned very shortly, the assumption was completely unfounded. The questionnaire provided the first shock. By the time we had completed six or seven drafts, and several pretests, the product (reproduced in Ap­ pendix A) was a substantial distance removed from the form of schedule used in GAF-I and GAF-II. (Despite considerable continuity of profes­ sional personnel, GAF-II had indeed differed from GAF-I in several im­ portant ways.) Although many of the same variables were included, the format and the phrasing were radically altered; for example, we intro­ duced a sharp change away from open-ended questions in the direction of precoded responses. In part this was a consequence of the somewhat different styles of questionnaire construction adopted by the Survey Re­ search Center, University of Michigan, the data collection agency for the 1955 and 1960 studies, and by National Analysts, Inc., of Philadelphia, the agency conducting the fieldwork for the 1965 study—although we must, of course, acknowledge that the final responsibility for modifica­ tions was entirely ours. We were continually confronted with trouble­ some choices between retaining the identical wording, for purposes of comparability, and trying to improve the measurement procedures— since there had been important developments in methodology, not the

CHAPTER I

least of which was the consequence of experience with aspects of GAF-I and GAF-II. It will soon become apparent to the reader that our choices have created problems. It is likely that our day-to-day involvement with the headaches caused by every little modification has led us to overstate the extent of the studies' noncomparability; in retrospect we are satisfied with the outcome of our attempts to provide the data required to continue the time series of many important reproductive phenomena from 1955 and 1960 through 1965. Nevertheless, we either seriously overestimated the degree of closure on methods of measuring central concepts or seriously under­ estimated our inclination or capacity for innovation. For example, as indicated in Chapters II and III, we developed strong reservations about the usefulness (for some purposes) of the measure of expected parity. In fact one of the principal departures of the 1965 National Fertility Study from its predecessors is that the latter were focused in part on the task of devising a variable for forecasting fertility—expected parity—as a way to complete the fertility histories of the younger and therefore in­ complete cohorts of United States women, for which Whelpton, with the assistance of Ryder and subsequently of Campbell, had prepared detailed tables from the entire range of official vital statistics and cen­ sus data.5 The 1965 National Fertility Study is more exclusively con­ cerned with the estimates of parameters than with the preparation of bases for forecasts, although the latter can scarcely fail to profit from the former. A further divergence in approach is evident in the chapters con­ taining analysis of fertility planning status, where new concepts and new approaches to measurement seemed to be highly desirable. However, in all cases of departure, we have attempted to provide for 1965 as accurate replicas as feasible of the basic measures reported for 1955 and 1960. REASONS FOR YET ANOTHER STUDY

It is certainly a legitimate question why another national fertility study should be required after each five-year interval. Many of the parameters estimated for the population in 1955, such as the incidence of sterility, are intrinsically unlikely to register much change over a period as short as a few years, and one would expect the same to be almost as true for relationships such as that between parity and socioeconomic status. A 5 See P. K. Whelpton and A. A. Campbell, "Fertility Tables for Birth Cohorts of American Women, Part 1," National OfBce of Vital Statistics, Vital Statistics/ Special Reports vol. 51, no. 1, January 29, 1960; and the extension of these tables for the years 1958-64 in the annual report. Vital Statistics of the United States, 1964, and for subsequent years in similar reports.

BACKGROUND

good defense of the legitimacy of such regular periodic surveys is of more than academic interest to us at this writing—since we are currently con­ ducting the 1970 Study. There are three types of reason for periodic repetition of essentially the same kind of survey—with the same underlying rationale as the regu­ lar decennial census: the measurement of trends, the assessment of the demographic effects of recent historical events, the continuation (and hopefully improvement) of theory and methodology. Many variables of demographic significance can and do change markedly within a five-year (or even shorter) interval, as the modern history of the crude birth rate in all developed nations clearly attests. Among the variables in which rapid change is not unlikely are: the pregnancy interval within which contraception is first used, the proportions of Catholics and nonwhites using contraception, the extent of reliance on surgical sterilization, the lengths of intervals between births, the success with which fertility is planned. To be more specific, consider the following developments which have occurred since the collection of data for the 1960 Study: the emergence of the birth control pill as the most important contraceptive now in use; the accelerating modification of attitudes toward and legislation concern­ ing abortion; the rocking of the Roman Catholic world by continuing controversy about contraceptive norms, most recently in response to the Papal Encyclical issued in July of 1968; the backdrop against which all of this has occurred, the most rapid sustained decline in the birth rate ever experienced in the United States, to a historic low of 17.4 in 1968. Each development has its reflection in the contents of this volume; each has presented us with its own peculiar methodological difficulties. Evalu­ ation of the pill's implications for the decline in the birth rate, estimation of the probability of discontinuing use of the pill, assessment of the pill's implications for coital frequency and time required for conception, analysis of the structure of attitudes toward abortion, and evaluation of the effects of the Papal Encyclical on the contraceptive practices of Cath­ olic couples—all of these have taken us in directions for which there was no precedent in the earlier studies. The Princeton Fertility Study was presented with a comparable opportunity in the late 1950's, that of as­ sessing the effects of an economic recession on fertility. Such natural so­ cial experiments represent some of the most important, yet unpredicta­ ble, payoffs from studies repeated over time. Finally, and this may be of the most enduring value, each study repre­ sents the opportunity (and responsibility) to make advances in substan-

CHAPTER I

tive theory and methodology, and thus add to knowledge of the nature of the fertility process and improve the measurement of the central con­ cepts of that process. Particularly from a methodological standpoint, the case for periodic surveys is crystal clear. Any single cross-sectional sam­ ple provides, in effect, a record of incomplete cohorts which, since their histories are differentially incomplete (truncated by interview at different stages of the family life cycle), are inherently incomparable to one an­ other. Furthermore, the representatives of each cohort are selected from that cohort on the basis of marriage at an age young enough to be in­ cluded in the sample (since the criteria for the sample include being mar­ ried). The biases implicit in these circumstances can only be removed by the collection of comparable information for these cohorts at succes­ sive stages in their reproductive histories. The ideal for such a purpose' is a longitudinal design, but, granted the expense and unique complexity of such an approach, the procedure of repeated cross-sections with the same basic format is a reasonable surrogate. These methodological prob­ lems beset any social scientist who uses the cross-sectional survey as his source of data; we demographers are most fortunate to have had the op­ portunity to solve the problem, through these periodic surveys, in a methodologically satisfying manner. SAMPLE DESIGN

The concept of the population to be sampled was defined as currently married women born since July 1, 1910, who were living with their husbands, residing in the coterminous United States, and able to partici­ pate in an English-language interview. The main differences in design be­ tween this sample and those of 1955 and 1960 relate to the inclusion of nonwhite women, the age limit, the number interviewed, and the time of year at which the interviews were conducted. The 1955 Study was confined to white women only, a decision prompted solely by budgetary considerations. The 1960 Study included a small sample (270) of nonwhites. In 1965 we greatly enlarged the nonwhite subsample by double-sampling Negroes, reaching a total of 1,201 women. The supplementary Negro sample was obtained by the preselection of household segments (ordinarily blocks) adjacent to those in the basic sample. If any Negroes resided in the adjacent segment, pre­ selected households in that segment were screened and all eligible Negro respondents so identified were designated for interview. Neither the 1960 nor the 1965 nonwhite sample is completely satis­ factory. The samples underrepresent young Negro couples at the lower

BACKGROUND

educational levels (as does the U.S. Census). In some respects the in­ adequacies raise serious questions about comparisons between the two samples: accordingly we exercise considerable caution in such cases and make liberal resort to qualifications. Even more serious than departures from the probable universe is the appropriateness of a nonwhite sample which is restricted to women currently married and living with their hus­ bands, since a significant fraction (more than one-third) of Negro fertility has occurred to women outside this category at the time of interview. Obviously all of the investigators in these studies have been aware of this limitation, yet all have felt frustrated by the logistical problems of col­ lecting reproductive and contraceptive histories from women who might, in fact, never have been married in the conventional sense of the term. In the 1970 National Fertility Study, we have defined the sample as women "ever married," rather than "currently married," in order to come still closer to the universe of interest. The age criteria of the three samples are different. In 1955, only women aged 18-39 were included. In 1960, those women who were 40-44 were added, to permit cohort observations for the women who were 35-39 in 1960; in fact, the bulk of the 1960 analysis was con­ centrated on the 18-39 age group. (An unsuccessful attempt was also made in 1960 to include women who would have been eligible for the 1955 but not for the 1960 study, because of change of marital status.) In 1965, we changed the criteria in several ways by defining the sample as women under 55 years of age or, more precisely, women born after July 1, 1910. First of all, this provided us with some women under 18 years old. Secondly, it gave us for the first time a sample of women aged 45-54. They were included for the purposes of learning more about the decline of fecundity in the later reproductive years and of studying once more, at the end of childbearing, cohorts which had almost the lowest fer­ tility in the history of the United States. Because we were less interested in this group than in the women of reproductive age, we drew only a halfsample of it. The analysis of these data is not routinely represented in this volume, although where appropriate (for example in the examina­ tion of attitudes toward abortion) the responses of the older women are included. In this case, as in the incorporation of Negro women in total sample estimates, appropriate weighting procedures have been employed. Most of our tables relate to two age segments—women under 45 or women 18-39. The former criterion is usually used when the interest is exclusively in the analysis of 1965 data; the 18-39 age category is em-

CHAPTER 1

ployed where comparisons with the earlier studies are involved, since this is the age range common to all three studies. Because of the addition of a double sample of Negroes and the expan­ sion of the age range, the size of our total sample (5,617) is appreciably larger than that of either GAF-I (2,713) or GAF-II (3,322). The size of the "core group" for comparisons across all thi^e studies—white women 18-39—is, however, fairly similar: 2,713 in 1955, 2,414 in 1960, and 2,919 in 1965. The following number of women were inter­ viewed in the different components of the 1965 sample: Total number interviewed

5,617

White women under 45 White women 45-54 Negro women under 45 Negro women 45-54 Other nonwhite women under 45 Other nonwhite women 45-54

3,769 647 969 158 72

2

The chief remaining difference in the sample design of the 1965 Study was the time of year at which the interviews were conducted. The central interview date for GAF-I was the beginning of March; for GAF-II, the interviews took place in May and June; and in 1965 the interviews were centered on mid-November. (A few scattered interviews were completed in 1966.) Since the age criteria for GAF-I and GAF-II were based on age at interview, while those for 1965 were based on age in mid-1965, women in the 1965 Study were several months older than their apparent counterparts in the earlier studies.6 These then were the main differences in sample design. Otherwise the samples in the three studies are, at least in theory, the same; each was an areal probability sample of the appropriate population living in the co­ terminous United States. Within the limits of sampling variability the estimates are presumably comparable. FIELDWORK RESULTS

In addition to error contributed by sampling variability, there is error introduced by the refusal of some women in the sample to be interviewed and by various other failures to conduct interviews with persons selected by the sampling procedure—the inability to find people at home, illness, and miscellaneous other categories. The interviewing completion rate in this study, defined as the number of successfully completed interviews 6 The

importance of this difference is explored further in Chapter ΠΙ.

BACKGROUND

(5,617) divided by the estimated number of women eligible to be inter­ viewed (6,397), was 88 percent. Of the 12 percent not interviewed, some two-thirds, or 8 percent, were classified as "refusals"; the remain­ der were cases of the respondent not at home despite calls, and miscel­ laneous other reasons. In Tables 1-1/1-4 we present some basic comparisons of our sample with estimates from the census data. Table I-I compares the age distribu­ tion of our white and Negro samples with that of the Current Population Survey (CPS) for 1965 and 1966. For the whites, the comparison is very close; for Negroes however, there appears to be an overrepresentation of those under age 25. Whereas the CPS estimates place about 22 per­ cent of the Negro population under age 25, the comparable figure in the NFS is 27 percent. Of course the Negro sample is much smaller than the white sample and therefore the amount of sampling error is greater, but this difference appears to exceed chance expectations. The sampling and various fieldwork procedures produced a larger Table 1-1.

Distribution of whites and Negroes from the 1965 National Fertility Study and Current Population Surveys for 1965 and 1966, by age·

CPS Age

March, 1965

NFS Nov., 1965

CPS March, 1966

White Women·(%) 14 - 19

4

4

4

20 - 24

16

17

25 - 29 30 - 34

19 19

17 20

17 19

35 - 44

42

41

19 42

100

100

100

Percent Total

Negro Women (%) 14 - 19 20 - 24 25 - 29 30 - 34 35 - 44 Percent Total

Sources:

4 18 20 20 39

6 21 19

5 17 18

19 36

20 39

100

100

100

U. S. Bureau of the Census, Current Population Reports P-20, no. 144 (November 10, 1965); and no. 159 (January 25, 1967).

CHAPTER I

overall nonwhite sample than was to be expected. The proportion of nonwhite women married, husband present, in the age group 14-44 was 9.4 percent according to the Current Population Survey estimates for both March 1965 and March 1966. The corresponding figure for the National Fertility Study is 12.9 percent, a considerable difference. (Other nonwhites as well as Negroes are overrepresented.) The apparent overrepresentation of nonwhites, relative to whites, is particularly evi­ dent in ages under 25. We have no satisfactory explanation for this discrepancy. Comparison of the educational attainment of women in our sample with that of the total population is severely handicapped by the absence of data on education by marital status in the Current Population Surveys. In Table 1-2 we compare the 1965 NFS distribution with that of women of comparable marital status in 1960. This shows a larger difference than actually exists, because of the strong intercohort trend toward more edu­ cation. For white and nonwhite women at every age, the NFS, compared with the 1960 Census, shows an underrepresentation of the lower and an overrepresentation of the higher educational categories. The differTable 1-2. Education of wives from Study,, by age and race.

Education

t^e

14·-44 1960 1965

I960 Census and the 1965 National Fertility

14--24 1960 1965

25-•34 1960 1965

35-44 1965 1960

White Women (%) College 4+ College 1-3 High School 4 High School 1 - 3 Grade School Percent Total

6 11 43 23 17

7 12 47 23 11

3 10 45 30 12

4 11 48 29 8

7 11 45 23 14

9 14 48 20 8

6 10 41 22 21

7 11 46 22 14

100

100

100

100

100

100

100

100

Nonwhite Women (%) Cbllege 4+ College 1-3 High School 4 High School 1 -3 Grade School Percent Total

Source:

4 6 25 29 36 100

30 35 23

2 6 28 39 25

2 10 36 45 6

5 7 27 30 31

6 8 34 30 21

4 5 20 24 47

4 6 21 32 37

100

100

100

100

100

100

100

4

8

U. S. Bureau of the Census, 1960 Census of Population. PC(2)-4E, Table 4.

BACKGROUND

ence is particularly strong in the lowest educational class, but this is the very category that has declined the most in recent years. In our judg­ ment, the educational level of the NFS sample is only a little higher than would be expected from census data. Perhaps the most important single respect in which it would be desir­ able for the sample to resemble the universe from which it is drawn, in a study such as ours, is the respondents' parity. Although the comparison is somewhat difficult to make, since the available data from official sources provide parity distributions for the marital status criterion of the study only for the 1960 Census (and a five-year difference in observation time is important for a dynamic variable like parity) and those data from the Current Population Survey for time points closer to the 1965 Study refer to ever-married women, we have, nevertheless, prepared Table 1-3 and believe it possible to make some inferences about our success in Table 1-3.

Mean parity for women married, husband present (M-HP), from the 1960 Census" and the 1965 National Fertility Study, and for ever-married women (E-M) from Current Population Surveys for 1960, 1964, and 1969, by age and race.

1960 Census Age

(M-HP)

1960 CPS (E-M)

1964

1965

CPS (E-M)

NFS (M-HP)

1969 CPS (E-M)

White Women 15 - 19 20 - 24

0.7 1.4

0.7

0.6

1.5

1.0

2.3

2.4

2.0

2.6

2.8

3.0

2.9

2.7

2.6

3.0

3.1

3.1

2.5

2.5

2.8

3.1

3.0

0.7

0.7

25 - 29

1.4 2.2

1.4 2.2

30 - 34

2.6

35 - 39

40 - 44

Nonwhite Women 15 20 25 30 35

-

19 24 29 34 39

40 - 44

Sources:

1.2 2.0 2.8 3.3 3.3

1.2 2.0 2.8 3.1 3.2

1.2 2.0 3.1 3.8 3.6

3.2

3.0

3.3

1.2 2.3· 3.3 '3.6 3.8 4.1

1.1 1.7 2.8 3.4 3.9 3.6

U. S. Bureau of the Census, 1960 Census of Population (M-HP), PC(2)-3A, Tables 16-17. U. S. Bureau of the Census, Current Population Reports Ρ-20» No. 178 (February 27, 1969).

CHAPTER I

this regard. The first two columns make it possible to gauge the extent to which reliance on data for ever-married women, rather than for women who are married, husband present, is likely to be a source of discrepancy. It is evident that only for the older age groups of nonwhite women is the marital status difference noteworthy. Accordingly we feel a fair degree of confidence in comparing the parity distributions of women by age and color for our study with the Current Population Survey results for March of 1964 and 1969, despite the difference in marital status criterion. For white women, the mean parity, age by age, is higher for our study than for the average of the preceding and succeeding Current Population Surveys. Examination of the detailed distributions (not shown) indicates that the principal source of discrepancy is an underrepresentation of in­ fertile women in our sample. A similar result occurs for the nonwhites. It is not surprising that women with children would be more likely to be found by our sample procedure than women without children, but, understandable though it may be, it remains a bias to be taken into ac­ count in the interpretation of our data. Although of minor importance for all but the youngest women, it should be noted for the sake of com­ pleteness that the women in each age group in the 1965 National Fertil­ ity Study are in fact some five months older (on the average) than their counterparts in the census and Current Population Survey sources, for reasons noted above. The last characteristic to be examined is the proportion of women in the sample who have married more than once. (See Table 1-4.) In the NFS sample of white women under 45 this proportion is 10 percent; for nonwhite women it is 18 percent. Comparing these proportions for women classified by age and education with the 1960 Census, we find an extremely close correspondence for white women. Although the patterns for nonwhites show correspondence, there are many more exceptions for them, and in general the percent remarried tends to run slightly higher in the nonwhite sample than in the 1960 Census. Here, and else­ where, the discrepancy may be attributable in part to errors in the census data as well as to the consequences of our sampling. THE INTERVIEW

In 1965 the staff of interviewers, and the sampling frames, were pro­ vided by National Analysts, Inc., a market and social science research organization in Philadelphia which had successfully undertaken that re-

BACKGROUND Table 1-4. Percent remarried for the 1960 Census and the 1965 National Fertility Study, by age, education, and race.

14-44 Education

1960

1965

Ik-24 1960

V

35·-44

25-•34

1965

1960

1965

1960

1965

White Women

Total

10

10

4

4

9

10

14

12

College 4+ College 1-3 High School 4 High School 1-3 Grade School

4 8 8 14 14

4 7 8 14 16

1 2 2 6 8

0 3 2 9 5

3 6 7 15 14

2 7 8 18 18

7 11 .12 18 16

8 10 10 14 19

Total

14

18

4

6

13

17

21

27

College 4+ College 1-3 High School 4 High School 1-3 Grade School

9 12 11 15 18

7 14 15 17 24

1 3 3 4 5

*

7 10 10 15 15

4 14 17 18 22

13 19 19 24 23

25 27 27 28

Nonwhite Women

7 3 8 *

*

* Base too small (N < 20). Source:

TJ. S. Bureau of the Census, 1960 Census of Population, PC(2)-4E, Table 5.

sponsibility in the longitudinal Princeton Fertility Study. They provided the services of a national staff of professional female interviewers whom we trained specifically for this study in two-day training sessions. The point of such training was to provide the interviewers with a good sense of the objectives of the study and the individual items, to acquaint them with the occasionally complicated routes through the interview schedule, and to prepare them for questions and difficulties that might arise in the interview situation. Some questions appeared somewhat sensitive, for example the one on coital frequency. It is our impression that if the in­ terviewer feels awkward or embarrassed about the question, it makes it much more difficult for the respondent to answer routinely. Part of our efforts were therefore devoted to defusing such questions by setting them in an antiseptic biological context. In general the interview schedule that we developed was quite straight­ forward and composed almost entirely of fixed-option questions. The questionnaire took an average of one hour and a quarter to administer, the main variation being due to the complexity of the respondent's

CHAPTER I

pregnancy history and, of course, the styles of both interviewer and respondent. The history of fertility studies in the United States has repeatedly dem­ onstrated that women are willing to give detailed personal information about their pregnancy and contraceptive histories, and the experience of the 1965 National Fertility Study is no exception. The reason for the continued success in eliciting this type of information is probably the fact that the subject matter is of such high interest to the woman. RELIABILITY OF INTERVIEW DATA

There is very little basis for evaluating the reliability of interview data without special methodological research. All that extensive editing of in­ terviews coming in from the field can accomplish—a procedure followed in all the studies—is to eliminate obvious incompetence or fabrication and correct clerical errors and some inconsistencies. This practice, al­ though essential, is unsuited to assessing the effects of differences in interviewing style or personality which could easily be a source of con­ siderable unreliability. Other sources of error, such as respondent variability, are also important. Many of the questions asked in these studies may simply not be sufficiently salient in the mind of the respond­ ent: although she will offer an opinion if asked a question, the underly­ ing attitude may be so uncrystallized that she may respond quite dif­ ferently if asked the same question another day. Such unreliability, of course, makes statistical prediction more difficult. And when the error variance introduced in the coding and data processing operations is added to interviewer and respondent variability, it is understandable why professionals in the survey research area are concerned with the problem. In order to help determine how much inconsistency is present in such studies, we reinterviewed with the same schedule 408 women randomly selected from the 5,617 initially interviewed. Analysis of these data will reflect the total amount of unreliability present in the measurement of many variables not subject to change, e.g., whether contraception had been used in a particular pregnancy interval in the past. Although some interviews were conducted by the same interviewers—thereby theo­ retically permitting the allocation of error to respondent and interviewer unreliability—this procedure was not developed systematically in the sense of randomizing the assignments. (It was a matter of economy and convenience in the field operation.) For purposes of methodological generalization, this reinterview feature of the 1965 National Fertility

BACKGROUND

Study is also limited by the arbitrary fact of a certain length of time be­ tween the two interviews—three months. Whether a longer or shorter interval would produce proportionately more or less error is difficult to determine. The results are discussed in Appendix A as a small contribu­ tion to this important methodological problem.7 7 For a comparable analysis of data obtained by reinterview, see C. F. Westoff et al., The Third Child, Appendix A.

CHAPTER II

Orientations Toward Numbers of Children SEVERAL types of que?';ons have been used in fertility surveys to deter­ mine the respondents' orientations toward numbers of children. Such data have been collected for a variety of related purposes ranging from population forecasting to analysis of the psychological aspects of fertility. In the 1965 National Fertility Study we attempted to measure four fertil­ ity variables: the number of children a woman intends to have; the number she expects to have; the number she would really desire·, and the number she considers ideal for the average American family. In this chapter, it is our purpose to examine some of the relationships among these responses with the objectives of determining the degree of re­ dundancy or substitutability and comparing other aspects of these different approaches.1 Our survey's predecessors, the Growth of American Families Studies of 1955 and 1960, collected data concerning wives' ideals and desires, on the one hand, and their expectations, on the other. One of our pur­ poses in designing the questionnaire for the 1965 Study was to attempt to sharpen still further the conceptual distinctions with respect to com­ pleted parity by asking the couple's reproductive intentions. We hoped that this would permit us both to separate the less realistic desire from the more realistic intent, and to measure directly (by comparing their intentions with their expectations) the extent to which the respondents perceive that the number intended might not in fact be achieved, because of inadequate reproductive control or capacity. INTENDED AND EXPECTED PARITY We define "total intended parity" as the sum of the respondent's current parity and the additional children she said she intended to have. "Current parity" is defined, for the purpose at hand, as the total number of previous live births to the respondent, plus one if the respondent was currently pregnant. Women who reported that they had passed meno­ pause or that they or their husbands had experienced a sterilizing opera­ tion were assigned a value of zero for the additional number of intended 1 This chapter was first published as Ν. B. Ryder and C. F. Westoff, "Relation­ ships Among Four Orientations to Total Number of Children: Intended, Expected, Desired and Ideal." Occasional publication of the Center for Population Research, National Institute of Child Health and Human Development, March, 1969.

CHAPTER Π

children. For the rest, the key set of questions to determine additional intended births was the following: Q. 44: (If you can have children) do you intend to have a(nother) child (after the one you are expecting now)? Those who answered "Yes" to Question 44 were asked: Q. 45: How many more children do you intend to have (not count­ ing your present pregnancy)? A nonnumerical response to Question 45 called for: Q. 46: Well, what is your best guess? Those women who did not answer Question 44 affirmatively were asked: Q. 47: Do you think you might later decide to have another child? 4 And if they said "Yes" to Question 47: Q. 48: How many more do you think you might decide to have? Women who said it was physically impossible for them to have a(nother) child (although they did not report that they had passed menopause or that they or their husbands had been sterilized) were asked: Q. 62: If it turns out that you can have a(nother) child, how many more children do you intend to have? These constituted 4 percent of our total sample of women under age 45 j approximately one-half of them gave a response other than zero to Ques­ tion 62. Such responses may be regarded as irresponsible, in the sense that their doubtful fecundity makes it unlikely that the women would have to substantiate their words with actions. On the other hand, there is likely to be an unrealistic quality to statements of intent on the part of many other women as well. In short, the number of children intended may be more a reflection of the state of the woman's mind regarding prospective childbearing than a forecast of her future fertility. In determining the additional number of children expected by the respondent, we assigned a value of zero to all women (some 17 percent of our sample) who reported, in response to a question concerning their reason for not using a contraceptive, that they were definitely sterile (in­ cluding menopause). All others were asked: Q. 124: We have now asked you questions about the number of children you would like to have, and the number you intend to have, about your physical ability to have children, and questions on family planning. Taking all this into account, how many more children do you expect to have (in addition to those you already have)?

ORIENTATIONS

If their response to this was nonnumerical, we asked: Q. 125: Many people feel the way you do, but still have some idea how it will turn out. As things are likely to work out for you, how many additional children do you expect to have? It is apparent that the form of question occasioned confusion for at least some women, because 38 percent of those who were currently pregnant gave a zero response to the question. It was no problem to recode the responses for these women, but we suspect that some others who were currently pregnant and gave non-zero responses to Questions 124-125 may also have misunderstood the question; we have no way of identify­ ing those errors. Table II-I has been prepared to permit a comparison of the results obtained from the questions concerning intended and expected numbers of children, by race, religion, and education of wife. (All analyses are restricted to the sample of women under 45.) The mean number of chil­ dren expected for each subset is expressed as the sum of current parity, additional children intended (Add. Int.), and the difference between the number of children expected and the number intended (Exp.-Int.). This form of presentation was chosen to display the relative magnitudes of three conceptually separable components of the total number of births expected. The largest source of variation is the number already born; smaller but still appreciable is the additional number intended; insig­ nificant are the variations in the residual difference between number expected and number intended. Although there are large differences by race, religion, and education in the total number expected, they are in no case attributable to a quantitative distinction between intention and expectation. The overall difference is 0.11 and the largest in the table is only 0.17. To see this in perspective, note that subsample sizes of the order of 2,000 are required to establish that a difference of 0.1 between means is statistically significant at the 5 percent level. Although few of the differences between mean intended and mean expected parity, shown in Table II-1, are statistically significant, they are not without pattern. If we calculate the proportion by which the mean additional expected exceeds the mean additional intended for the three education groups, we find the excess is 9 percent for those with at least some college, 21 percent for those who have completed high school, and 37 percent for those who have not completed high school. These relative differences at the margin hold up systematically within each of the racereligion groups. Nevertheless, the outstanding observation is that the

CHAPTER U Table II-l. Components of total expected parity (current parity, additional intended and difference between expected and intended) by race and religion, by education of wife.

Total Education

Components

White Women NonCatholic Catholic

Negro Women Total

Expected Parity Total

Expected Current Add. Int. Exp.-Int.

3.36 2.77 0.48 0.11

3.06 2.56 0.41 0.09

3.83 3.00 0.70 0.13

3.28 2.69 0.49 0.10

3.92 3.34 0.45 0.13

College

Expected Current Add. Int. Exp.-Int.

2.99 2.24 0.69 0.06

2.78 2.12 0.59 0.07

3.81 2.78 0.99 0.04

3.00 2.26 0.67 0.07

2.86 2.14 ι 0.68 3.04

High School 4

Expected Current Add. Int. Exp.-Int.

3.16 2.53 0.52 0.11

2.86 2.35 0.43 0.08

3.76 2.88 0.71 0.17

3.13 2.51 0.52 0.10

3.40 2.68 0.62 0.10

Cohort

20-24 25-29 30-34 35-39 40-44

20-24 25-29 30-34· 35-39 40-44

Under 25 1916-20 1921-25 1926-30 1931-35 1936-40

3.14 3.36 3.40

3.08 3.32 3.50

2,.97 3,.06 3,.41

3.14 3.26

3,.16 3,.35 3,.59

3.46 3.54

646 567 611

651 560 671

560 573 646

448 667

242 247 323

190 312

318 326 323

258 355

Under 20 1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

3,,20 3..07 3,,26

3.32 3.46 3.53

3.32 3.49 3.58

307 311 413

320 304 345'

297 285 375

20 - 24 1916-20 1921-25 1926-30 1931-35 1936-40

2.97 3.24 3.24

2.88 3.15 3.41

2,.82 2,.83 3,,22

2.91 3.01 346 263 266

354 275 296

For the means of those married before age 25, the 95 percent confidence limits are approximately i 0.14; for the means of the two age-at-marriage subgroups, they are approximately i 0.22. ^ Estimates for 1955, 1960, and 1965 appear on the upper, middle, and lower diagonals respectively.

aged 25-29, we obtain 3.2 and 3.3, a reversal of the previous difference. An age-at-marriage control is itself, however, an incomplete solution, because it poses the new problem of the extent to which the expectations of members of a cohort who marry before age 20 are predictive of ex­ pectations of those who marry after age 20. Those who marry young are selected on characteristics, such as lower education, which are related to high fertility (because of ineffective contraception); moreover, they are exposed to the risk of accident longer and are less likely to en­ counter subfecundity before they achieve their intended final parity. There is no guarantee, furthermore, that the consequences of selectivity by age at marriage will remain the same during a period when age at marriage is changing.

TREND OF EXPECTED PARITY

The second problem of intercohort comparison from a single survey is that any two cohorts are by definition at different ages at the same time. The straightforward solution to this problem is to use the results of two surveys and make each intercohort comparison for the same inter­ view i£ge, but this possibility was not open to the GAF-I investigators. On the basis of their expectations data they produced forecasts of the eventual mean parity of all women.9 For the birth cohorts of 1926-30 (C. 1926-30) they predicted 2.86; this now appears to be approximately 9 percent too low. For C. 1931-35 they predicted 2.90; this now appears to be approximately 16 percent too low.10 Clearly expectations estimates from a single survey may be unreliable forecasts. In GAF-II the investigators were in a much stronger position to manage some of these problems, because they could compare the ex­ pected parities of the same cohort at two successive ages. Change would be expected to occur not only because individual couples modify their own expectations, but also because the composition of the cohort is somewhat transformed through time, in ways that are probably related to expected parity, by the process of marital dissolution and remarriage. There are also problems in the use of different samples of the same ag­ gregate, with somewhat different questionnaires, coding practices, and the like. The problem of the same cohort's expectations changing over time deserves special attention because it is directly relevant to the use­ fulness of these data for projection. A comparison of expected parity with eventual parity would give a misleading appearance of relatively accurate expectations, because most of the total births expected by most cohorts would have already oc­ curred. To emphasize that part of expected parity which is in the future, we show in Table III-2 the additional numbers of births expected by cohorts in 1955 and in 1960, and the absolute and relative intracohort changes in the subsequent time intervals. The following observations seem pertinent. (1) There has been substantial intracohort change in 9 It is evident to anyone who has glanced at the forecasting chapters of GAF-I and GAF-II that, even if a successful forecast could be made of the final parities of the various cohorts of white married women, husband present, much additional estimation is required to achieve forecasts of the eventual mean parity of all wom­ en. In particular, estimates of final parity are required for all white women not in the married, husband-present category at time of interview, and for all nonwhite women. 10 The cohort fertility tables to which reference is made from time to time in this chapter are described in: Ν. B. Ryder, "The Emergence of a Modern Fertility Pattern: United States, 1917-66," in S. J. Behrman, Leslie Corsa, Jr., and R. Freedman, eds., Fertility and Family Planning (Ann Arbor: University of Michigan Press, 1969), pp. 99-123.

0.31 0.64 1.33

0.17 0.41 0.86 1.70

0.26 0.54 1.U

Add'l Exp.

40.09 -0.05 40.18

40.30 40.03 40.17 +0.26

40.17 -0.02 40.18

4- 29 - 8 4- 14

4-177 4- 7 4- 20 4- 15

4- 65 - 4 4- 16

Change Percent in Exp. Change

1955-60

0.27 0.60 1.22

0.20 0.42 0.68 1.37

0.25 0.51 0.93

Add'l 'Exp.

40.08 -0.03 40.07

40.09 0 40.02 40.36

40.10 -0.01 40.04

Change in Exp.

1960-65

4- 30 - 5 4- 6

4- 45 0 4- 3 4- 26

4-40 - 2 4- 4

Percent Change

0.64 1.33

0.41 0.86 1.70

0.54 1.11

Add'l Exp.

40.03 40.15

40.12 40.17 40.28

40.08 40.17

4- 5 4- 11

+ 29 4- 20 + 16

4- 15 4- 15

Change Percent in Exp,· Change

1955-65

Additional expected births and absolute and relative changes in expectations, for white cohorts 1916-40, between 1955-60, 1960-65,*»nd 1955-65,* by age at marriage.

* For the purposes of these calculations, the expected parities in 1965 have been reduced by 0.1, to compensate for the bias noted in the text.

1916-20 1921-25 1926-30 1931-35

20 - 24

1916-20 1921-25 1926-30 1931-35 1936-40

Under 20

1916-20 1921-25 1926-30 1931-35

Under 25

Cohort

Age at Marriage

Table III-2.

TREND OF EXPECTED PARITY

expectations. (2) Although increases predominate, the amounts of change, both absolute and relative, are highly variable from cohort to cohort and time period to time period. This variability is a considerable embarrassment for the person wishing to use expectations as forecasts. (3) The discrepancies do not seem to differ systematically by age at mar­ riage or period of observation. The investigators in GAF-II were well aware of these problems and accordingly introduced into their fertility forecasts an allowance for the systematic increase of expected parity with age, although they recognized that the irregularity of expectations changes makes it difficult to apply adjustments with confidence. Their adjusted forecast for the 1931-35 cohort was 3.30; it now appears that this will be approximately 2 percent too low; their forecast for the 1936-40 cohort was 3.14; it now appears that this will be more than 5 percent too low. Clearly the 1960 fore­ casts are superior to those of 1955, in large part because the expecta­ tions data were not accepted at face value. The principal basis for their forecast that mean parity would decline from 3.30 for C. 1931-35 to 3.14 for C. 1936-40 was apparently the cir­ cumstance that those aged 25-29 in 1960 expected 3.4 while those aged 18-24 expected only 3.0. With the advantage of our 1965 data, and the assistance of a control for age at marriage, we see in the middle panel of Table III-I how they may have been led astray. For women married be­ fore age 20, C. 1931-35 observed at age 25-29 expected 3.46, and C. 1936-40 observed at age 20-24 expected 3.07. But if these two cohort groups are compared at the same interview age, their expected parities are 3.46 and 3.53. From this one would infer that the direction of change was upward rather than downward. What can we now say about the movement of expected parity from C. 1931-35 to C. 1936-40? For those married before age 25, and inter­ viewed at ages 25-29, the mean for C. 1931-35 was 3.36 and that for C. 1936-40 was 3.40. Taking into account the probable upward bias of approximately 3 percent in the 1965 data, a small decline from the earlier to the later cohort would appear likely, but such an inference be­ comes tenuous indeed when considered in the light of possible changes in expected parity between age 25-29 and the end of the childbearing period, as implied in the data shown in Table III-2. Although one may raise the objection that the variations being examined are too small to be statistically significant, they are nevertheless not too small to be demographically significant. Otherwise stated, the changes from cohort to cohort which we need to be able to measure for the purpose of making

CHAPTER III

fertility projections in a usefully narrow range may necessitate samples of much larger size than those on which we are reporting here. Evaluation of the trend from C. 1936-40 to C. 1941-45 is much more difficult, because at present we are restricted to a comparison of the parities expected by those married before age 20. The mean expected parity is 3.07 for the earlier and 3.26 for the later cohort. Given the small positive bias to the latter figure, which is taken from the 1965 Study, we conclude that there has been little change, although this con­ clusion must be qualified by the considerations that women who marry young are unreliable predictors of the behavior of the members of their cohort who marry at later ages, and that we are now dealing with smaller sample sizes (a confidence interval of ±0.22). It is probably evident that we view expectations forecasting with little enthusiasm. Since the other writings on this subject (see footnote 4) con­ vey a contrary view, it is pertinent to examine the kinds of evidence they present. One source of support has been the alleged stability of expecta­ tions from period to period. Although it is true that expected parity dis­ tributions of women in the childbearing ages for successive periods re­ semble each other closely, one can easily be deceived by such a compari­ son. When fertility is rising or falling, the proportion in each intermediate parity is increased by transfers from one contiguous parity and decreased by transfers into the other; variability is concentrated in the tails of the distribution. Furthermore, successive cohorts tend to have similar parity distributions because they share so much common history. And even if this were not so, the comparison of distributions for the entire 18-39 age range has the effect of a moving average which tends to erase instability. There may be at present a fortuitous counterbalancing of increased ex­ pectations because of accidental conceptions (and increased intentions) for some women, and decreased expectations because of unanticipated sterility (and decreased intentions) for others, but this is not helpful to the cohort projection problem if the counterbalancing tends to occur between, rather than within, cohorts. For example, the respondents in GAF-I expected 0.75 births in the next five years, and the follow-up of the same cohorts in GAF-II showed 0.80 actually born, an error of only +7 percent. But this result was the outcome of counterbalancing intercohort inaccuracies. The error was jT16 percent for C. 1916-20, -j-12 percent for C. 1921-25, —8 percent for C. 1926-30 and —6 percent for C. 1931-37. Perhaps the strongest psychological support for expectations as fore­ casts came from the GAF-II prediction that total cohort fertility may

TREND OF EXPECTED PARITY

have started to decline, coincident with the dramatic drop in the birth rate. Period-specific mean parity fell by 25 percent in 5 years, from 3.62 in 1961 to 2.73 in 1966. This is the kind of change that expecta­ tions data should help us to foresee, and it is what the investigators in GAF-II thought their data showed them. (They predicted a decline from 3.65 for 1955-60 to 3.53 ± 0.30 for 1960-65; the actual average for the latter period was 3.38.) Yet we have already seen the weaknesses of their interpretation and come to a somewhat different conclusion about the cohort trend on the basis of our 1965 data. Their prediction of de­ cline appears to have been right for the wrong reasons.11 The rapid decline of period fertility during the 1960's may signify a like change in cohort fertility, or it may only reflect a change in the direc­ tion of change in the timing pattern of cohort fertility, or (most likely) some combination of these two influences. We wish it were possible to resolve this question with our expected parity data, but we do not really believe it is. Our data suggest no change in the eventual total, but the wide band of error implicit in the necessary sequence of inferences sets the alternative possibilities somewhere between a small decline and a small rise. Nevertheless, since our caution would be unlikely to prevent others from making inferences from these data, we feel obliged to come up with the guess that there is now occurring a large rise in the age of childbearing combined with a small decline in eventual mean parity. We are confident that there has been some tendency for women to bear their children later, and we believe that later is likely to mean fewer. Delayed childbearing reduces exposure to the risk of accident and increases exposure to the risk of subfecundity; it gives couples the opportunity to scale down intentions as a response to the exposure to alternatives to fertility; it gives them experience with contraception which they may exploit to terminate their fertility at the intended level. It is also our im­ pression that unintended births will decline with the rise in education, the improvement of contraceptive methods, the diminution of Roman Catho­ lic resistance to effective means of fertility regulation (despite the Papal Encyclical), and the increase in government assistance to family planning. 11 Vital statistics released since this book was first prepared for publication, and revised procedures for estimating fertility of cohorts still in the childbearing ages, suggest that the GAF-II estimates for C.1931-35 and C.1936-40 will turn out to be essentially correct. In other words, the authors of GAF-II seem to have done better with 1960 data and no age-at-marriage control than we could with 1965 data and the refinement possible with that control. In the text we charge them with being right for the wrong reasons; we should now add that we were ap­ parently wrong for the right reasons.

CHAPTER III

Nevertheless, total fertility seems unlikely to decline by much more than 10 percent for couples of C. 1941-45 relative to those of C. 1936-40. The key problem for the forecaster will remain what it has been for the past 30 years: the time pattern of cohort fertility. Our knowledge of the reasons why couples had so many children at young ages during the 1950's is scanty, vague, and impressionistic. Indeed, the new pattern of somewhat later fertility among the most recent cohorts seems a much more understandable mode of behavior. It is essential to have projections of the time pattern, as well as the level, of cohort fertility, because the purpose of fertility forecasts is to produce annual estimates of births, and this requires knowledge of both dimensions of cohort reproductive behavior. Although surveys of married women do provide some important timing data (as discussed in the following sec­ tion), they are intrinsically ill-designed to provide information on what may be the most important determinant of the time pattern of fertility— the time pattern of marriage. Even more problematic for the success of a period-by-period fore­ cast is the circumstance that each year a new cohort moves into the re­ productive age span and begins to make even the most recent survey of married women out of date. For example, in 1967 more than 30 per­ cent of the babies born in the United States were born to the cohorts of 1946 and later, about whom our survey provides no usable information; by 1970 they constituted the majority of the childbearers. Indeed, if our concern about the validity of using the expected parity of women married before age 20 to represent the behavior of all women inhibits us from employing C. 1941-45 data for projective purposes, then we are faced with the circumstance that more than 60 percent of the births in 1967 occurred to the cohorts of 1941 and later. Our conclusion on the value of expected parity as a way of determin­ ing the current trend in cohort fertility, and thus of improving fertility forecasts, is that it is a small and untrustworthy addition to our stock of projection procedures. We are not now taking a position concerning the value of expected parity for other purposes, such as differential fertility analysis (see Chapter IV). It is obviously convenient to have some sur­ rogate for final parity when dealing with differentially incomplete cohorts, rather than to strain sample size with controls for marital dura­ tion, age at marriage, and the like, even if the surrogate is to some extent a piece of information about the respondent's state of mind at the time of interview.

TREND OF EXPECTED PARITY

THE TIME PATTERN OF FERTILITY Our data do permit some observations to be made concerning the timing as well as the level of cohort fertility. In Table III-3 we show, for cohorts and interview ages available from the three surveys, the propor­ tion of those married by age 25 who were married by age 20. In conTable III-3.

Proportion (per thousand) of those married by age 25 who were mrried by age 20 for white cohorts 1916-40: 1955, 1960, 1965.a

Age at Interview Cohort 1916-20 1921-25 1926-30 1931-35 1936-40

a

25-29

481 536 565

30-34

456 509 559

35-39

40-44

432 431 500

424 468

Estimates for 1955, 1960, and 1965 appear on the upper, middle, and lower diagonals respectively.

sidering the trend of age at marriage revealed by a table like this, we must remember that there is no information concerning those first mar­ ried subsequent to age 25, about those married before age 25 but not in the married, husband-present category at time of interview, or about those who remain single. It is also evident from the intracohort varia­ tions in these proportions that small changes must be regarded with caution. Nevertheless, the strong impression is that there has been a major in­ crease in early marriages throughout the range of experience reviewed here. These data, however, are not pertinent to the recent decline in young marriages—Current Population Reports (P-20, No. 159) for Jan­ uary 25, 1967, gives the median age for females at first marriage as 20.2 in 1958 and 20.6 in 1965—because the years in which these five cohort groups reached their median marriage age were approximately 1938, 1943, 1948, 1953, and 1958, respectively. The cohort of interest for recent nuptiality changes is C. 1941-45, and its youngest members reached age 25 only in 1970. We may also draw from Table III-I an indication of the difference that age at marriage makes for expected parity, as shown in Table III-4.

CHAPTER III 5able IXI-4. Difference between expected parity of those married before and those married after age 20, for white cohorts 1916-40: 1955, 1960, 1965.a

Age at Interview Cohort 1916-20 1921-25 1926-30 1931-35 1936-40

a

25-29

0.35 0.22 0.26

30-34

0.44 0.34 0.17

35-39

40-44

0.34 0.52 0.37

0.55 0.53

Estimates for 1955, 1960, and 1965 appear on the upper, middle, and lower diagonals respectively.

It is clear that expected parity is markedly higher for those married be­ fore age 20 (by an average of 11 percent over the five cohort groups). This difference, however, is not entirely real. The younger the age at which those of the same cohort were married, the longer they have been married at the same interview age. (The difference in marital duration is approximately 3.5 years.) Consequently, they have had longer ex­ posure to the possibility of having their expected parity unintentionally increased by an accidental birth. Nevertheless, an appreciable margin in favor of young marriage would remain even if comparisons were ad­ justed for this bias. A comparison of the differences in expected parity for cohorts at the same interview age in 1960 and 1965 shows for each succeeding cohort except C. 1936-40 a smaller differential by age at marriage than its predecessor. Such a development would be a not unlikely consequence of the declining selectivity of the young-marriage group as progressively larger proportions of the total cohort enter that category (as seen in Table III-3). Increasing proportions married at younger ages and per­ sistently higher expected parity for younger than for older marriages have combined to increase expected fertility for the cohorts in this series. We also have data on the timing of childbearing within marriage. Table III-5 shows current parity as a proportion of expected parity. In interpreting this table, it is important to keep in mind the strictures noted above concerning the probable increase of each cohort's expected parity with age. Furthermore, the level of current parity as of 1965, particu­ larly for the younger cohorts, is affected by the circumstance that the

TREND OF EXPECTED PARITY Table III-5. Eatio (per thousand) of current to expected parity for white cohorts 19X6-40, by age at marriage and interview age: 1955, 1960, 1965,a

Age at Marriage Cohort

Af;e at Interview 20-24

25-29

30-34

35-39

40-44

915 920 953

967 989

946 939 970

977 993

889 903 934

959 984

Under 25 1916-20 1921-25 1926-30 1931-35 1936-40

649 723 734

827 847 901

Under 20 1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

469 554 597

742 805 826

877 881 939

20 - 24 1916-20 1921-25 1926-30 1931-35 1936-40

a

552 623 605

778 808 850

Estimates for 1955, 1960, and 1965 appear on the upper, middle, and lower diagonals respectively.

women in each stated age interval in 1965 were 4.5 months older than those in the same age interval in 1960. By age 25-29 the cohorts of 1926-30, 1931-35, and 1936-40 had achieved 65 percent, 72 percent, and 73 percent, respectively, of their expected parity. This suggests a decline in the age distribution of childbearing. An even more sensitive index of the timing of childbearing is the proportion of remaining expected fertility achieved within the next 5 years. Between ages 25-29 and 30-34, C. 1926-30 achieved 56 percent of its remaining fertility; and C. 1931-35, 64 percent. Similarly, between ages 30-34 and 35-39, C. 1921-25 achieved 54 percent of its remaining fertility and C. 1926-30, 69 percent. Finally, between ages 35-39 and 40-44, C. 1916-20 achieved 61 percent of its remaining fertility and

CHAPTER III

C. 1921-25, 86 percent. This is impressive evidence that the acceleration of the tempo of childbearing between 1955-60 and 1960-65 has been occurring through the entire age span. Table III-5 also permits some comparisons of the timing of marital fertility by age at marriage, but these comparisons turn out to be much less readily interpretable than they might at first appear. Although those who married younger in each cohort have achieved a higher proportion of their expected parity (when compared at the same interview age) than those who married older, the principal reason is the difference in marital duration. Those married at younger ages have by time of interview been married 3.5 years longer than those married at older ages. Accordingly, it does not seem feasible to use these data to compare the timing of marital fertility by age at marriage. It is worth noting, however, that the later cohorts, because a higher proportion of their members married be­ fore age 20, have somewhat longer marital durations than the earlier cohorts, and this is one explanation for the rise in the proportions of expected parity that have already been achieved (Table III-5). SUMMARY

In this chapter we have presented data on the expected parity of white women from three national-sample surveys in 1955, 1960, and 1965. On the basis of details in which the 1965 survey is not precisely comparable with GAF-I and GAF-II, it is estimated that there is a positive bias of approximately 3 percent in expected parity in 1965 relative to 1960. An age-at-marriage control has been systematically employed throughout the chapter in comparing the expected parity of women in different age groups at the time of interview. Because of this, and the advantages of hindsight, some of the findings of GAF-I and GAF-II have been reinterpreted. In particular, their forecasting attempts appear to have been complicated by the considerable and highly variable tendency for a cohort's expectations to increase with age. We have reviewed the various kinds of evidence concerning the use­ fulness of expectations data for fertility forecasts and conclude that their value is quite limited. It appears to us, however, that cohorts now in the early years of marriage, as compared with their immediate predecessors, will experience a substantial rise in the age of childbearing combined with a small decline in eventual mean parity. This rise in the age of child­ bearing would represent a sharp reversal of the trends prevailing during the 1950's when American women were both marrying and completing larger families at progressively earlier ages.

CHAPTER IV

Racial, Religious, and Socioeconomic Differences in Fertility ONE of the main traditional interests in fertility research is the study of group differences in fertility. Quite aside from its utility for such policy considerations as the assessment of family planning needs or the projec­ tion of population change on the basis of subgroup trends, the chief objective of differential fertility research is to understand the sociocultural determinants of fertility. This chapter is organized around three such perspectives: white-Negro differences; religious differences (pri­ marily Protestant-Catholic) and trends in the religious differential since 1955; and socioeconomic differences in fertility and trends. The main index of fertility employed is the total number of children expected. As a surrogate for (in the form of a partial forecast of) final parity, this index enables us to circumvent the difficulties inherent in comparisons of the childbearing of differentially incomplete cohorts. Expected fertility was selected not because of any particular analytical advantage over, say, intended fertility (see Chapters II and III), but mainly because it facilitates comparison with the 1955 and 1960 data. Some use is also made of the number of children desired. THE WHITE-NEGRO DIFFERENTIAL Expected and Desired Family Size

By the time of the 1965 interview, white women under 45 years of age had borne an average of 2.6 births; by comparison, the fertility of Negro women was 3.3 births, a difference of 0.7 births (Table IV-I). Project­ ing the completed fertility of these women on the basis of their replies to questions about the total number of children expected, the difference appears to be maintained at an average of 3.3 for whites and 4.0 for Negroes. The additional Negro fertility has already occurred. In other words, Negroes expect more children than whites because they have already had more, as shown in Table IV-I; this occurs despite their de­ sire for fewer children. We observe from the comparative values for the number of children desired that white and Negro women seem to prefer the same number of children on the average—their mean desired parities are 3.3 and 3.2 respectively. Since the number of children desired (and the number ex-

2.9

3.5

High School 1-3

Grade School

4.8

3.5

2.6

2.3

1.4

3.3

3.1

2.9 3.2 4.0

3.2

3.3

3.8 4.1

3.4

2.9

2.7

3.2

3.0

3.3

W

5.2

4.1

3.4

3.1

2.3

4.0

N

Total Expected

.25

.13

-.12

-.32

-.24

-.06

1.23

.83

.50

.43

-.49

.72

.54

.54

.70

.72

.90

.66

.44

.60

.79

.74

.92

.64

N

W

W

N

Additional Expected

Expected Minus Desired

396

850

1790

457

277

3770

W

212

343

301

74

38

968

N

Number of Women

Numbers constitute basis for calculation of mean current parity. Because of more nonresponses, the wean numbers of children desired and expected are based on slightly lower figures.

2.4

High School 4

a

2.2

College 1-3

2.8

3.3

2.1

College 4+

3.2

3.3

2.6

N

W

N

W

Total

Education

Total Desired

Current Parity

Table IV-I. Mean current parity, -total number of children desired, total number expected and additional number expected by white and Negro women.

SOCIOECONOMIC DIFFERENCES IN FERTILITY

pected) by white women is the same as the number desired by Negro women, and the number desired by Negro women is significantly lower than the number they expect, the obvious conclusion is that the higher fertility of Negro women is due to unsuccessful planning. Another way of expressing the same pattern of relationships is to say that, if fertility reflected only the wishes of the parents, there would be little if any differ­ ence between the two groups. (A further confirmation of this inference is that, among women classified as having completely planned their fer­ tility, the total number of children expected by white and Negro women is very similar.) Education Both the number of children desired and the number expected are clearly related to the educational attainment of the women interviewed (Table IV-I). For both races there is an inverse relationship between the two fertility indices and the amount of education. This is especially striking for the number of children expected by Negro women; the aver­ age number expected by Negroes with less than any high school educa­ tion (5.2) is more than double the average number expected (2.3) by college graduates. The association is much stronger with expected than with desired fertility because the differential ability to limit children is much more significant for the former than for the latter measure. In other words, the less education a woman has received, the more the num­ ber of children she expects exceeds the number she desires. We cannot, however, ignore the fact that, although the association is not as strong as with the number expected, the number of children de­ sired also tends to vary inversely with education. Can we conclude, therefore, that women with little schooling actually want more children than those with more education? These women have already borne vary­ ing numbers of children, and, despite the logical niceties of the questions about the number of children wanted—which ask the respondent whether she would "just as soon have had fewer"—there is probably a strong tendency for women to report the number of children they already have as wanted. One admittedly imperfect way to approach this question is to examine the relationship between education and the number of children desired among women of a given parity. (The approach is im­ perfect because, as parity increases, the incidence of excess fertility in­ creases; the latter, in turn, is subject to rationalization about whether or not the last child was wanted.) Since for white women above the grade-

CHAPTER IV

school level there is no association at all between education and the num­ ber of children desired, the question exists mainly for Negroes. Table IV-2 indicates a definite contraction of the relationship when parity is controlled, but the negative correlation between education and family-size preferences persists nonetheless. Some of this persistence, Table IV-2.

Number of children desired by Negro women, by education and parity.

Education

Total

0

1

3

2

5+

4

Number of Children Desired Total High School 4+ High School 1-3 Grade School

3.2

2.6

2.5

2.7

3.2

3.1

4.3

2.8 3.2 4.0

2.6 2.6 2.8

2.4 2.6 2.9

2.5 3.0 2.7

3.0 3.1 3.6

3.0 3.1 3.2

3:7 4.2 4.7

Numbfer of Women Total High School 4+ High School 1-3 Grade School

959

117

151

171

163

99

258

411 338 210

61 36 20

93 35 23

88 61 22

67 69 27

45 39 15

57 98 103

though not all, is probably due to rationalization of excess fertility by women in the grade-school category. A series of differences between white and Negro fertility is presented in Table IV-3. The excess of Negro over white fertility is sharply de­ lineated in current parity and the total number expected. As can be seen Table IV-3. Differentials (Negro minus white) in current parity, total number of children desired and expected, and additional number expected.

Current Total ParityDesired Total College 4+ College 1-3 High School 4 High School 1-3 Grade School

Total Expected

Expected Minus Desired

Additional Expected

.73

-.07

.71

.78

-.02

-.76 .17 .21 .57 1.24

-.49 -.56 -.32 -.07 .16

-.74 .19 .30 .63 1.14

-.25 .75 .62 .70 .98

.02 .02 .09 .06 -.10

SOCIOECONOMIC DIFFERENCES IN FERTILITY

from the lack of relationship between education and the racial differen­ tial in the additional number of children expected (last column), the difference for current parity accounts for most of the difference for the total number expected. The racial differential for the number desired also varies with education, but most of the differences are negative—that is, Negroes at all levels of education except grade school express prefer­ ences for fewer children than whites despite the fact that aside from col­ lege graduates they already have more children. Finally, if we examine the racial difference between the total expected and the total desired— which can be considered as a measure of past or anticipated excess fer­ tility—we find an inverse relationship with education. Age, Duration of Marriage, and Age at Marriage In this section we pursue the variation of the racial differential by age, duration of marriage, and the age at which the woman (first) married (Table IV-4). The racial differential in the total number of children ex­ pected varies directly with the woman's age. The source of this variation is probably some combination of a trend (reflected by younger women) toward more education, and the associated tendency to plan fertility more successfully, with some greater tendency for young Negro women to underestimate their actual future fertility. A similar though less regular relationship obtains for the racial differ­ ential in desired family size except that in the youngest group—under 25—Negroes want fewer children than whites. The racial differential in the excess of the number of children expected over the number desired varies directly with age. All of this would seem to point toward the con­ vergence of fertility patterns for the two races in the near future. The same patterns are evident in the racial differential by the duration of marriage; the longer the women have been married the greater the excess of Negro over white fertility indices. Age at (first) marriage is of course simply the difference between cur­ rent age and the number of years since the (first) marriage. The racial differential in the number of children expected by age at marriage is quite strong, ranging from a full child among those married before age 18 to a quarter of a child for women married after age 24. The earlier exposure of Negro women to the risk of pregnancy (with obvious implications for future fertility) appears to be the explanation. (Negro women married before age 16 expect an average of 5.4 children, compared with 4.0 for white women in the same category.) The racial differential in desired family size also varies with age at marriage, ranging from virtually zero

Under 18 18 - 19 20 - 21 22 - 24 25+

Age at Marriage

Under 5 5 - 9 10 - 14 15 - 19 20 - 24 25+

Number of Years Married

Under 20 20 - 24 25 - 29 30 - 34 35 - 39 40 - 44

Ase of Wife

Total

Variable

N

3.6 3.3 3.2 3.1 2.5

2.9 3.3 3.3 3.4 3.2 3.8

2.8 3.1 3.3 3.5 3.3 3.1

4.6 4.0 3.6 3.7 2.8

2.9 3.7 4.5 4.4 4.3 •5.2

3.0 3.4 4.0 4.4 4.1 4.2

3.3 4.0

W

1.00 .73 .40 .59 .27

.01 . .47 1.18 .99 1.06 1.46

.17 .32 .69 .97 .76 1.10

.71

N-W

Number Expected N

3.4 3.3 3.3 3.3 3.3

3.0 3.2 3.3 3.4 3.5 3.7

2.9 3.1 3.2 3.3 3.4 3.5

3;3 3.3 3.1 3.1 2.6

2.7 2.9 3.6 3.3 3.7 4.1

2.6 2.8 3.3 3.5 3.2 3.6

3.3 3.2

W

-.05 .08 -.14 -.20 -.68

-.36 -.31 .30 -.06 .16 .39

-.25 -.34 .06 .19 -.23 .15

-i07

N-W

Number Desired

.35 .64 .78 .89 .90 .57

.42 .66 .63 .78 .99 .95

.78

.26 1.22 1.05 -.02 .63 .65 -.10 .44 .54 -.17 .62 .79 -.79 .16 .95

-.11 .26 .37 .06 .84 .88 -.03 .85 .88 -.02 1.03 1.05 -.30 .60 .90 .08 1.15 1.07

-.07 -.02 .15 .11 -.09 -.38

.72

Expected Minus Desired N-W W N

843 1060 888 571 288

686 .744 733 826 514 160

154 625 630 735 719 801

3664

282 256 179 124 75

228 193 189 159 109 62

54 200 177 179 165 165

940

Number of Women W N

Table IV-4. Number of children.expected and desired by white and Negro women, by age of wife, number of years married, and age at (first) marriage.

t O σ*

SOCIOECONOMIC DIFFERENCES IN FERTILITY

for women who married under 20, and increasing only slightly for those married between 20 and 25, to a considerable difference (0.68) for those who married at older ages. For both races, the difference between the number of children expected and the number desired varies inversely with age at (first) marriage. Socioeconomic Status

We have already observed that the amount of education a woman has received bears a pronounced relationship to the number of children she expects (Table IV-I). The fact that her husband's occupational status and the amount of income he receives bear a similar relationship is there­ fore not surprising (Table IV-5). The relationship is especially striking for the Negroes, whose average number of children expected ranges from 2.8 in the upper white-collar class to 6.0 in the farm group. The relation­ ship is also negative for white women, but the variation is much less than for Negroes. As a consequence, the racial differential (Negro minus white) also varies inversely with the husband's occupational class, rang­ ing from —.22 children in the upper white-collar class to +.75 in the lower blue-collar class and -)-2.40 in the farm group. The source of the high fertility expectations of Negroes in farm occupations presumably extends beyond the status dimension to encompass the whole complex of Southern rural culture, as discussed below. In all nonfarm occupations, the number of children desired by Negro women is lower than that desired by whites, although the racial differen­ tial is essentially nonexistent in the lower blue-collar class. The number expected and the number desired are fairly consistent for the different occupational strata in the white population and for upper white-collar Negroes, but Negro women in the other occupational classes expect from about 0.5 to 1.5 more children than they say they want. The negative association of the husband's income with expected fertil­ ity also covers a wide range among Negroes, from an average of 2.6 chil­ dren expected by women whose husbands earn $10,000 or more to 5.1 children expected by women below the conventionally defined poverty line of $3,000. An even sharper relationship with family income (which includes wife's earnings) prevails among Negroes: expected parity ranges from 2.1 for families with incomes of $15,000 and more (com­ pared with 3.0 for comparable white families) to 5.3 for families below the poverty line (appreciably higher than the 3.9 children expected by poor white women). Some of this racial difference among poor couples is associated with the Southern and rural-farm differential, but some also

N

3.3 4.0

W

white-collar white-collar blue-collar blue-collar

and over - 9,999 - 7,999 - 6,999 - 5,999 - 4,999 - 3,999 - 3,000

$15,000 and over 10,000 - 14,999 8,000 - 9,999 7,000 - 7,999 6,000 - 6,999 5,000- 5,999 4,000 - 4,999 3,000 - 3,999 0 - 3,000

Family Income

$10,000 8,000 7,000 6,000 5,000 4,000 3,000 0

Husband's Income

Upper Lower Upper Lower Farm

3.0 3.0 3.3 3.3 3.3 3.4 3.4 3.5 3.9

3.1 3.2 3.4 3.1 3.3 3.3 3.3 3.5

3.1 3.0 3.3 3.5 3.6

2.1 2.6 3.2 3.5 3.6 3.9 4.3 4.7 5.3

2.6 3.5 3.3 3.2 3.7 3.5 4.3 5.1

2.8 3.4 3.5 4.2 6.0

-.91 -.43 -.06 .21 .31 .48 .83 1.13 1.33

-.56 .31 -.07 .04 .40 .23 1.00 1.58

-.22 .36 .15 .75 2.40

.71

N-W

Number Expected

Husband's Occupational Class

Total

Variable N

3.4 3.2 3.3 3.3 3.3 3.3 3.4 3.4 3.7

3.3 3.3 3.4 3.3 3.3 3.3 3.3 3.5

3.3 3.3 3.3 3.4 3.7

2.7 2.8 3.0 2.8 3.0 3.1 3.3 3.6 3.9

2.9 3.3 2.8 3.0 3.2 2.9 3.5 3.7

2.8 2.8 3.0 3.3 4.5

3.3 3.2

W

-.73 -.40 -.38 -.44 -.30 -.18 -.01 .20 .16

-.44 .08 -.60 -.34 -.12 -.43 .19 .15

-.49 -.52 -.31 -.01 .81

-.07

N-W

Number Desired

.72

N .78

N-W

-.34 -.24 -.05 -.03 .03 .11 .09 .12 .24

-.52 -.18 -.27 -.03 .27 .32 .62 .65 .64 .61 .77 .66 .93 .84 1.05 .93 1.41 1.17

-.18 -.30 -.12 -.08 .17 .23 .01 .54 .53 -.16 .23 .38 .03 .55 .52 .00 .66 .66 .04 .85 .81 .00 1.43 1.43

-.19 .08 .27 -.24 .63 .88 -.01 .45 .46 .12 .88 .76 -.03 1.56 1.59

-.06

W

Expected Minus Desired

274 715 632 416 421 441 263 143 123

722 502 398 524 596 366 229 215

1030 481 902 962 166

3664

W

22 63 77 74 114 153 137 129 112

27 28 51 86 184 185 182 183

65 95 179 516 54

940

N

Number of Women

Table IV-5. Number of children expected and desired by white and Negro women, by husband's occupational class, husband's income, and family income.

SOCIOECONOMIC DIFFERENCES IN FERTILITY

reflects the educational and other cultural deprivations of Negroes in urban ghettos. Because the association of income with expected fertility is so much weaker for whites, the racial differential varies sharply with income, especially family income. It is significant that the tabulations of desired family size by race and family income reveal that Negro women in all but the two lowest income classes express preferences for fewer children than white women. In the poorest classes they indicate only slightly higher desires—i.e., 0.2 more—whereas in all but the top three family income groups their expectations are higher. As a consequence, there is a sharp negative correlation between the racial differential in ex­ pected minus desired family size and income. Wife's Employment

The principal reason that family income shows a sharper, more con­ sistent association with fertility than the husband's income is that the main additional component of family income is money earned by the wife. Since the working wife would be likely to have fewer children and also increase the family income, the combination operates to augment the negative association. The lower expected fertility of the working wife can be seen in Table IV-6. Even women who are not currently employed but who have worked sometime since marriage expect and desire fewer children than those who have never worked after marriage. The strength of this association appears to increase with the wife's age (see lower panels of Table IV-6). The longer the woman has been married, the greater the opportunity she has had to be employed, and the more time during which other selective factors have had a chance to operate. The racial differential in the number of children expected tends to be the greatest for women who have never worked and smallest for women currently working. A different pattern prevails for the number of chil­ dren desired. The racial differential is virtually nonexistent except for women who are now working. Negro women in the labor force, espe­ cially those working primarily for reasons other than simple financial need, desire fewer children than white women in the same categories. Among white women currently working, the number of children ex­ pected is less than the number desired; among Negro women in all cate­ gories except those who like to work the number expected is greater than the number desired, with the greatest disparity occurring among those who have never worked. The relationship between fertility and the wife's work history is diffi­ cult to interpret not only because of the age factor but also and primarily

3.2 3.2 2.8 2.7 3.0 2.6

3.8 3.4 3.0 3.3 2.9 2.9

3.9 3.1 2.9 3.1 3.0 2.4

Never Worked Once Worked Now Working: For family needs For extra things Likes to work

Never Worked Once Worked Now Working: For family needs For extra things Likes to work

Never Worked Once Worked Now Working: For family needs For extra things Likes to work

*

5.5 3.8 3.7 4.5 2.9

*

4.8 4.5 3.7 4.2 3.3

*

3.4 3.6 3.0 3.5 2.7

4.5 4.0 3.6 4.3 3.0 2.4

*Base too small (N < 20).

3.7 3.3 2.9 3.1 2.9 2.6

3.1 3.1 3.0 3.1 3.0 3.0

.29 .39 .20 .71 -.25

*

3.7 3.4 3.4 3.5 3.5 3.2

1.68 .65 .86 1.44 -.11

ft

3.5 3.2 3.1 3.2 2.9 3.3

1.00 1.08 .65 .90 .42

ft

3.5 3.3 3.2 3.3 3.2 3.2

*

3.9 3.5 3.2 3.2 3.1

ft

3.9 3.5 3.0 3.1 2.8

*

2.7 3.0 2.6 2.6 2.5

3.5 3.3 3.0 3.1 2.8 2.8

*

.23 .10 -.22 -.25 -.40

35-44

*

.42 .27 -.14 -.06 -.10

25 - 34

ft

-.37 -.11 -.49 -.51 -.30

Under 25

.03 .07 -.25 -.19 -.37 -.35

Under 45

Number Desired N N-W W

.88 .76 .67 1.19 .06 -.19

Number Expected N-W N W

Never Worked Once Worked Now Working: For family needs For extra things Likes to work

Wife's Work History

*

.15 1.60 1.45 -.21 ,34 .55 -.53 .55 1.08 -.39 1.30 1.69 -.46 -.17 .29 * -.77 *

ft

.58 .81 .79 .96 .52

ft

.71 .66 .59 .50 .43 .69 .84 1.22 .22 .05

.32 .90 .15 .96 -.09 .70 .07 1.03 -.05 .47 -.37 *

.05 .09 -.26 -.38 -.04 -.39

.18 1.03 .85 -.01 .70 .77 -.34 .58 .92 -.24 1.14 1.43 -.24 .19 .30 -.60 -.44 .41

Expected Minus Desired W N N-W

356 615 549 184 174 171

296 711 358 138 123 88

246 300 233 99 80 47

898 1626 1140 421 377 306

73 81 176 104 52 17

82 122 152 69 62 19

88 86 80 28 41 10

243 289 408 201 155 46

Number of Women W N

Table IV-6. Number of children expected and desired by white and Negro women, by age and wife's work history.

SOCIOECONOMIC DIFFERENCES IN FERTILITY

because of the interrelations between fecundity and employment. Un­ doubtedly many women wish to work and therefore deliberately restrict their fertility. Other women have low fertility (for whatever reason) and therefore have the freedom to work. Some working women become preg­ nant and have to stop working. And still other women work because they have had more children than can be supported by their husband's in­ come. This problem of inferring causal sequences can only be resolved satisfactorily in a longitudinal study. On the basis of our data and those from similar studies, all that can be safely asserted is that working wives expect fewer children, whatever may be the causal sequence. Women who were working at the time of the interview were asked their reasons for working in the following question: "What is the main reason you are now working—is it because the family really needs the income, because you want to get extra things for yourself and the family, or because you like it?" The analysis of the consequences for fertility of different reasons for working is not obviously affected by the selective factors affecting whether women work, although it is by no means totally free of complications; for example, many women working for family needs are doing so because of excess fertility. The respondent who is working because she "likes to work," the closest to the "career woman" type, shows the lowest fertility for both whites and Negroes. The fertility expectations of white and Negro women who work for reasons other than family needs are very similar, although the number of children desired by Negro women runs lower than for white women. The "career woman" is a considerably more com­ mon phenomenon among whites than among Negroes; she represents 27 percent of white working wives but only 11 percent of Negro working wives. This could be due to differences in both educational backgrounds and employment opportunities, to the greater economic need for Negro women to work, and perhaps to cultural differences in the meaning of work which could affect the expression of reasons. Among both white and Negro career women however, the number of children desired ex­ ceeds the number expected, a difference that may arise from either infecundity or the sacrifices a woman who pursues a career may have to make in her family life. Residence As anticipated, the highest fertility expectations for Negro women are expressed by those living in the South; in all other regions the racial dif­ ferential approaches equality (see Table IV-7). In fact, if we exclude the

5.7

3.6

A

Central cities of 14 largest metropolitan 3.2 areas Suburbs of largest 3.2 metropolitan areas Central cities 150,000 - 1,500,000 3.2 Central cities 3.5 50,000 - 150,000 Suburbs of cities 3.1 50,000 - 1,500,000 Smaller cities and towns3.1 3.4 Rural nonfarm 3.8 Rural,farm

2.11

1.07

.60 .69 1.52 1.10 .88 1.96

3.8

4.2

4.6 4.2 4.2 5.7

*

4.1

3.0

*

A

it

3.8

.78

-.14

4.0

3.3

.32

-.02 .31 1.26 -.43

.71

N-W

3.1

3.5

3.3 3.7 4.3 2.9

3.2

3.3 3.4 3.0 3.3

4.0

3.3

Size of Place of Residence

Never lived on farm Once lived on nonsouthem ,farm Now lives on nonsouthern farm Once lived on southern farm Now lives on southern farm

Farm-Resion Residence

Northeast Midwest South Far West

Region of Residence

Total

N

W

Number Expected

3.4 4.3

3.2 3.6

*

3.0 3.3 3.0 3.2 -3.8 4.3

3.3 3.2 3.5 3.3 3.3 3.4 3.6

2.8

*

3.6

3.3

3.1

3.0

3.4

3.3

2.9 3.2 3.4 2.8

3.2

3.3

3.2 3.4 3.2 3.4

N

W

N-W

Number Desired

81

892

.94 1.04

-.16 1.62 1.78 -.17 1.02 1.19 -.01 .41 .42 .19 1.42 1.23

-.26 -.09 .46 .73

.84 .84

.00 -.15

.72 .73

*

817 549 535 275

277

82 87 98 88

69

241

291

258 17

304

.30 .25

89

294

--

170 135 590 45

616

122

.05 1.42 1.37

*

377

.93

.72

-.21

164

*

.01

-.12

*

-.44

.74

.14

476

2109

.62

.57

842 1130 1071 621

.39 .31 .45 .44 .94 1.13 .13 .26

940

3664

.78

N

.72

W

Number of Women

N-W

N

*

.26

-.10

-.26 *

-.05

.08 .01 -.19 -.13

-.30

-.33 -.13 .13 -.69

-.06

W

Expected Minus Desired

I O

Characteristic

Table IV-7. Number of children expected and desired by white and Negro women, by region of residence, history of farm residence, and size of place of residence.

I ' »-Γ ! ο Γ -r O Γ

SOCIOECONOMIC DIFFERENCES IN FERTILITY

South, the fertility expectations of Negroes correspond closely with those of whites (3.4 and 3.3 respectively). In all regions except the South, Negroes express preferences for fewer children than do whites; even in the South the racial differential is negligible. The difference be­ tween the number of children expected and the number desired shows little variation for whites, whereas for Negroes there is an excess of ex­ pected over desired parity in all regions, especially in the South. A similar and not unconnected generalization can be offered about the effects of farm residence on Negro fertility. The highest fertility indi­ cated—and the largest racial differential—is found among Negro women now living on a southern farm. This group represents the least modern­ ized segment of the Negro population, with the lowest education and the least exposure to the values of the modern world. It also reflects the vestiges of a culture in which high fertility was intrinsic to the economic and kinship systems. The highest expected fertility among white couples is also evident for those currently living on a farm, with little apparent difference between regions. The effects of farm residence on Negro fertility are still visible even after a couple has left the farm for the city. Thus the average number of children expected by Negro couples who once lived on a farm but who are now residents of towns or cities is intermediate between that of couples still on a farm and that of couples who have no history of farm residence. Why this effect does not appear for the expected fertility of whites is not clear. In any event, the primary consideration in these comparisons is the fact that the expected fertility of Negroes with no farm residence is very similar to that of whites and, since rural life is rapidly becoming a thing of the past, the rapid disappearance of racial differences in fertility seems imminent. This prospect is also suggested by consideration of the racial differential in the number of children desired. Only among women who now live or who once lived on a southern farm does the number desired by Negroes exceed that desired by whites. And among those Negro women currently living on Southern farms, the number of children ex­ pected considerably exceeds the number desired, more than in any other group. There is also more variation in Negro than in white expected fertility by size of the place of residence. The racial differential in expected fertility is clearly greater in less-populated areas and is even reversed in the central cities of the nation's largest metropolitan areas. (These varia­ tions are highly correlated with education.) It is particularly noteworthy

CHAPTER IV

that Negro women desire fewer children than do white women in all urban areas of the country; only in rural sections do Negroes express preferences for larger families. In all sizes of community, the fertility expectations of Negro women exceed their desires. Trends in the Racial Differential

In the 1960 Study, only a very small sample of nonwhites (270) was included. Because of that small number but also because of other prob­ lems which raise questions about the comparability of the 1960 and 1965 samples of nonwhites, only a few observations about trends in the racial differential will be offered. For purposes of comparison, the groups con­ sidered are white and nonwhite women 18-39, rather than white and Negro women under 45. The overwhelming majority of nonwhites in both samples are, of course, Negroes.1 Our interest here is directed to the question of whether the difference between white and nonwhite expected fertility has contracted between 1960 and 1965. If we look at the total number of children expected at the two dates for the entire age range, we get the impression that there has been little change in the differential (Table IV-8). Slightness of the change is due to the stability of the differential in the South. It is possible that some expansion of the differential may have occurred in the Mid­ west, but small numbers inhibit generalization. There is no particular pattern to the trend in the racial differential by the wife's education. In the 1965 sample white and nonwhite women who had attended college appear to expect the same number of children. Some increase in the differential appears to have occurred in the highschool graduate group, but differentials for the remaining (lower) cate­ gories show little change. In general, education seems to have less of an effect on the racial differential in 1965 than in 1960. Again it should be emphasized that these detailed statistics are frequently based on very small numbers (especially in the 1960 sample) and that many of the apparent changes may consequently reflect nothing more than sampling variability. THE RELIGIOUS DIFFERENTIAL

In this part of the chapter we are concerned with the differences in fertility among white women of different religions, particularly the dif­ ference between Protestants and Catholics. We are also interested in the 1 In the 1960 sample 94 percent of the nonwhites were Negroes; in 1965 it was 93 percent.

3.0 3.0 3.3 3.7

3.2 3.3 2.9 3.2

3.1

Base too small (N < 20).

College High School 4 High School 1-3 Grade School

Wife's Education

Northeast Midwest South Far West

Region of Residence

Total

Characteristic

2.4 2.9 3.8 4.7

*

2.9 3.4 4.0

3.6

3.0 3.1 3.5 4.2

3.4 3.4 3.0 3.3

3.3

2.9 3.5 4.2 4.9

3.5 4.1 4.2 3.5

4.0

-.6 -.1 .5 1.0

*

-.3 .1 1.1

.5

-.1 .4 .7 .7

.1 .7 1.2 .2

.7

Nvtmber of Children Expected and Desired 1960 1965 NW -W 1960 1965 W NW W NW

427 1153 579 255

581 715 744 374

2414

37 73 86 74

41 74 136 19

270

571 1377 627 256

650 870 827 485

2832

105 282 280 150

156 128 489 45

818

Number of Women 1960 1965 m Nff W W

Table IV-8. Nvimber of children expected and desired by white and nonwhite women 18-39, by region of residence, and wife's education, for the 1960 and 1965 Studies.

CHAPTER IV

social and economic factors modifying these differences, and in the clos­ ing sections we focus on the trend in the religious differential. The difference in fertility between Protestants and Catholics bears some resemblance to that between whites and Negroes, but the compari­ son is more apparent than real. Catholic women have experienced higher fertility thus far (an average of 0.5 more births) and expect 0.9 more children on the average than Protestant women (see Table IV-9). The source of the difference between religions is, however, radically different Table IV-9.

Current parity and the mean number of children expected and desired, by religion of wife.

Number of Children

Number of Women

Prot.

Cath.

Jewish

Prot.

Cath. Jewish

Mean Current Parity

2.3

2.8

2.1

1908

846

56

Mean Number Expected

3.0

3.9

2.9

1864

809

55

Mean Number Desired

3.0

3.7

3.0

1893

834

56

Ratio Expected to Desired

1.00

1.04

.95

from that between the races. We observed earlier that the average num­ ber of children desired by Negro women was the same as that desired by white women, and that the higher fertility expected by Negroes was to be explained by differences in fertility planning and control rather than by different norms. (See Chapters V and IX.) The average number of children desired by Catholic women (3.7), on the other hand, is not very different from the average number expected (3.9), suggesting the hy­ pothesis that, unlike Negroes, the high fertility of Catholics is in large measure a function of their family-size preferences. These averages sug­ gest that Catholics' expectations are slightly higher than their desires. Some 27 percent of Catholic women expect more children than they want, compared with 20 percent of Protestant women (Table IV-14). The proportion expecting fewer than desired is identical for the two groups—22 percent. In sum, Catholics expect more children in part be­ cause they already have more children but also because they want more children. These variables are all to some extent confounded both by their inter­ actions with each other (see Chapter II) and by the processes of accept-

SOCIOECONOMIC DIFFERENCES IN FERTILITY

ance and rationalization. Thus, one reason why Catholics have more children than Protestants is because they want more, but there is un­ doubtedly some tendency for women to report the number they have already borne as wanted, despite our asking the respondent whether she would "just as soon have had fewer." One indication of this is the fact that about half of the women who reported an unwanted birth also said that they did not prefer fewer children than they had. To examine the effect of differences in achieved parity on the differ­ ences in family size desired by Protestants and Catholics, we prepared Table IV-10. Among women at the same parity, Catholics desire more children than Protestants. We must conclude therefore, that the process of adjusting desires to fertility is not the chief explanation of ProtestantCatholic differences in family-size preferences. The same type of analy­ sis is repeated for the number of children expected (second panel of Table IV-10). It shows the same results; regardless of current parity, Catholic women expect to have more children than Protestant women. Because the numbers of children expected and desired are so much more closely related to each other among Protestant and Catholic white Table IV-10. Number of children desired and expected, by wife's religion and parity.

Parity Religion

2

Total

Number Desired Protestant Catholic C - P

3.0 3.7 .69

2.7 3.4

.62

2.6 3.1 .57

2.7 3.2 .53

3.2 3.8 .65

3.8 4.0

3.3 3.7 .42

4.2 4.7 .48

394 173

223 124

.18

Number Expected Protestant Catholic C - P

3.0 3.9

1.7

2.2

2.6

2.9 .71

.83

2.5 2.9 .49

Number of Women Protestant Catholic

3

1864 809

242

330

116

511 183

Includes women desiring or expecting more than four children.

^ Numbers constitute bases for calculation of mean number of children expected. Because of fewer nonresponses, the average number of children desired is based on slightly larger numbers.

CHAPTER IV

women than among white and Negro women, our analysis of the religious differential in this chapter rests exclusively on the total number of chil­ dren expected and does not follow the practice, as in the preceding sec­ tion, of carrying along parallel series of desired and expected family size. As other studies have consistently shown, the fertility of Jewish women is the lowest of the three major religious groups, although it is not markedly different from that of Protestants and is, as we shall see, the same as that of several leading Protestant denominations. The small number of Jewish respondents in our sample precludes any detailed analysis of their behavior. Our data permit us to examine the implications for fertility of Protestant-Catholic intermarriages. Both types of such intermarriage are much closer in levels of expected fertility to Protestant than to Catholic marriages. The combination of Protestant wife and Catholic husband shows an expected parity value of 3.1, slightly lower than the average of 3.3 children expected by Catholic wife and Protestant husband. (The average for Protestant marriages is 3.0; that for Catholic marriages is 4.0.) This pattern probably reflects in part the greater influence of the wife's religion on fertility and in part the probability that Catholics in­ volved in mixed marriages are drawn disproportionately from individuals less religiously involved than those in all-Catholic marriages. (Half of the Catholic women married to Protestant men attend mass weekly or more, compared with three-quarters of the Catholic women married to Catholic men.) The fertility of Protestant women, classified by their denominational preference, is shown in Table IV-11. Except for women belonging to Fundamentalist sects and, to a lesser extent, Lutherans, the two measures of fertility show a striking similarity across denominations. There is so little variance in fact that the validity of the classification "Protestant" is strengthened considerably. Religiousness

In our interview we asked a series of questions designed to measure the couple's involvement in religious activities from which we con­ structed an eight-item index of religiousness for Protestants and Catho­ lics separately. Six items are common for both religions: whether the couple were married at a religious ceremony, the frequency with which the wife sees a clergyman about personal problems, whether there are daily family religious activities, how religious-minded the woman regards herself, whether she attended religious instruction classes, and whether

SOCIOECONOMIC DIFFERENCES IN FERTILITY

Table IV-11. Nvimber of births and number of children expected by Protestant women, by denomination.

Mean Births

Denomination Total Protestant Fundamentalist Sects Baptist Lutheran Episcopalian Methodist Presbyterian Congregationalist Evangelical Other Protestants

a

Number3 of Women

Mean Total Expected 3.0 3.4 2.9 3.2 2.9 2.9 2.9 2.9 2.8 3.2

2.3 2.8 2.3 2.5 2.1 2.2 2.2 2.2 2.2 2.4

1908 219 516 228 90 402 162 38 36 196

The base number is for live births; the numbers for total expected fertility are slightly lower because of nonresponses.

as a child she participated in family religious activities. The two dissimi­ lar items are, for Protestants, the frequency with which the wife attends religious services and the frequency with which she prays at home; and for Catholics, the frequency with which the respondent attends Mass and the frequency with which she receives Communion. Each item was dichotomized and the scores simply added, producing an index of re­ ligiousness ranging from a low of 0 to a high of 8. The association of this index with the number of children expected is shown in Table IV-12. There is no evidence of any relationship between religiousness and ex­ pected fertility among Protestant women. For Catholic women, however, Table IV-12. Number of children expected by Protestant and Catholic women, by an index of the wife's religiousness.

Index of Religiousness

Mean Number Expected Protestant Catholic

Number of Women Protestant Catholic

Total

3.0

3.9

1864

807

Low 0-1 2 3 4 56 7 High 8

2.9 3.0 3.1 3.1 3.0 3.0 2.9 3.1

3.6 3.3 3.6 3.6 3.9 4.1 4.5 4.7

171 288. 354 345 279 248 137 42

41 69 108 128 156 154 112 39

CHAPTER IV

the association (except for one deviant case), is striking, The average of 3.9 children expected by all Catholic women is a combination of a range of 3.3 to 3.6 children expected by women of less than average religious­ ness and a range of 3.9 to 4.7 children expected by women of more than average religiousness, the final figure applying to women classified as "religious" on all eight items. For the most part, the items comprising the index of religiousness tap the extent of formal and informal associational involvement in religious culture. We also asked two questions designed to measure the couple's community involvement in religion, the answers to which are reported in Table IV-13. One question aimed at measuring the religious composition of the couple's neighborhood. Since the variable of religiousness seems relevant to fertility only for Catholics, the pertinent question is: "Of the people who live in your neighborhood about how many are Catholic? Would you say: none, less than half, half, more than half, all?" The dis­ tribution of Catholics by the religious composition of the neighborhood is indicated in the appropriate "Number of Women" column. The rela­ tion of this variable to expected fertility only shows some slight but irregular pattern in the predicted direction. The highest value for the number of children expected is 4.1 for the Catholic women Uving in predominantly Catholic neighborhoods; the mean is slightly less for the rest. The result may reflect self-selection of neighborhood or perhaps bias in the response. The second approach to the communal dimension of religiousness is contained in the question: "Of the friends you feel close to, about how many are Catholic? Would you say: none, less than half, half, more than half, all?" The distribution of responses to this question reveals that most Catholic women report half or more of their closest friends are Catholic. Only a small fraction—4 percent—reported no close Catholic friends compared with 14 percent claiming that all of their close friends are Catholic. The association of this variable with expected fertility is strong­ er than that of neighborhood religious composition. The average number of children expected ranges from 3.3 to 3.5 for CathoUc women of whose close friends less than half are of the same religion, to 4.1 and 4.2 chil­ dren expected by women most of whose friends are Catholic. The relationships between religious homogeneity and the individual's expected family size, whether measured for neighborhoods or for friend­ ship groups, involves a problem of interpreting the causal sequence. Peo­ ple with particular interests and values, in this instance of a reUgious nature, tend to associate with people with similar orientations, and such

Number of Children Expected

3.9

3.9

3.8

3.9

4.1

4.1

Total

Less than Half

Half

More than Half

All

4.2 ) > 4.1(

3.8

3.5

3.3

3.9

3.1

3.0

3.0

3.0

3.0

3.0

3.2

2.9

3.0

3.0

Catholics Protestants Neighborhood Friends Neighborhood Friends

None

Proportion Catholic

62

220

176

248

49

755

111

278

224

165

29

807

7

153

213

708

649

1730

9

99

294

818

642

1862

Catholics Protestants Neighborhood Friends Neighborhood Friends

Number of Women

Table IV-X3. Number of children expected by Catholic and Protestant women, by the proportion of neighborhood and close friends that are Catholic.

CHAPTER IV

an environment undoubtedly operates to reinforce those interests and values. Thus the higher fertility orientations of Catholic persons in reli­ giously homogeneous social environments have to be interpreted both as a selective process and as an effect. Our study is not designed to evaluate the relative significance of the two mechanisms. We also examined the fertility behavior of Protestant women in terms of the extent of their involvement with Catholics in both neighborhood and friendship contexts on the premise that the higher fertility culture Cr Catholics might have some effect on Protestant fertility values. However, no such influence can be discerned; the expected fertility of Protestant women seems completely unrelated to the extent of their social involve­ ment with Catholics (Table IV-13). Education and Religion

The proportion of women who expect to have more children than they want is closely connected with education, and the relationship is stronger for Protestants than for Catholics. Among Protestants the percentage of women expecting more children than they want increases sharply as edu­ cation declines, from a low of 10 percent among college graduates to a high of 35 percent among women with eight or fewer years of schooling (Table IV-14). Presumably this reflects the importance of differences in education for fertility planning. Table IV-14. Percent of wives who expect fewer children, the same number, or more than they desire, by religion and education.

Expect Fewer than Desire

Expect Same as Desire

Expect More than Desire

Percent Total

Number of Women

Protestants Total College 4 College. 1-3 H. S. 4 H. S. 1-3 Grade School

22 29 21 24 19 21

57 62 65 59 52 44

20 10 14 18 29 35

100 100 100 100 100 100

1836 147 236 878 421 154

Catholics Total College 4 College 1-3 H. S. 4 H. S. 1-3 Grade School

22 18 35 20 20 25

51 62 42 52 53 42

27 20 23 28 27 33

100 100 100 100 100 100

785 39 86 407 169 84

SOCIOECONOMIC DIFFERENCES IN FERTILITY

For Protestant women there is a sharp inverse relationship between the number of children expected and the wife's educational attainment, ranging from 2.7 children among college women to 3.9 children among women at the grade school level (Table IV-15). The pattern of the edu­ cational differential among Catholic women, however, is not the same as among Protestants. As shown by many other analyses in this study, the Catholic women with four years of higher education are a distinctive Srigh-fertility group, expecting nearly as many children as women with only a grade school education and more than women at intermediate levels of education. The reason for this exception is that Catholic women graduating from college are drawn disproportionately from the more devout Catholic population with high fertility norms. It is in this group, interestingly enough, that we find the highest proportion (62 percent) of Catholic women who expect to have the same number of children that they want. The comparably high fertility expectations of the least edu­ cated segment of the Catholic population are due more to ineffective fertility planning than to a high fertility norm. The relationship between Protestant and Catholic fertility at each educational level and the association between education and fertility for younger and older women of each religious group is also shown in Table IV-15. The largest Protestant-Catholic difference in the average number Table IV-15,. Number Of children expected by Protestant and Catholic women, by .wife's age and education.

18 - 29

18 - 39 Education

P

C

C-P

P

C

30 - 39 C-P

P

C

C-P

3.1 2.7 2.8 2.9 3.4 4.0

4.0 5.0 3.6 3.9 3.8 4.9

.98 2.32 .80 1.04 .42 .96

941 88 110 455 186 102

424 24 49 215 83 53

Number of Children Expected Total College 4+ College 1-3 High School 4 High School 1-3 Grade School

3.0 2.7 2.7 2.8 3.4 3.9

3.9 4.5 3.5 3.8 3.8 4.7

.88 1.83 .78 .96 .45 .74

3.0 2.8 2.6 2.8 3.3 3.9

3.7 3.9 3.5 3.6 3.8 4.3

.76 1.15 .83 .88 .49 .39

Number of Women Total College 4+ College 1-3 High School.4 High School 1-3 Grade School

1863 154 239 890 425 155

809 42 88 418 173 88

922 66 129 435 239 53

385 18 39 203 90 35

CHAPTER IV

of children expected is found among women with at least four years of college education, a difference that is especially pronounced among the older women. Among younger women (18-29) there appears to be a positive relationship between education and the excess of Catholic over Protestant fertility, but among older women (30-39) there is no sys­ tematic pattern below the college graduate level. In general, the levels of expected fertility for the younger and older women parallel each other fairly closely, with slightly lower values among the younger women. (For a discussion of this age difference, see Chapter III.) We have also looked at the relationships of the education of the hus­ band and the education of the couple with the number of children ex­ pected by the wife (results not shown); no particular surprises are revealed. The education of the husband varies with expected fertility in the same direction as does the wife's education, but the association is considerably weaker. Use of a classification of both husband and wife for education does not alter the patterns of association at all, but it does show the greater influence of the wife's education. In a separate analysis of data from the 1965 National Fertility Study, Robert M. Dinkel examined the kinds of influence that the education of the parental generation has upon the number of children desired by the younger generation.2 He found that, when the educations of the spouses and the wife's parents are on the same relative3 level, the result is to re­ inforce the typical class fertility values of the wife, and in so doing to widen class differences. This "widening" is, however, very slight; in gen­ eral, the education of the husband and that of the wife's parents do not explain much additional variance of desired fertility above that ac­ counted for by the wife's education. In still another independent analysis of the relationship between edu­ cation and fertility,4 Larry Bumpass demonstrated the importance of the wife's age at marriage. The negative association of education with fertil­ ity is seen to weaken or even reverse for the higher age-at-marriage categories—a phenomenon observed for non-Catholics as well as Catho­ lics. Bumpass speculates that both social factors (the incidence of nonfamilial adult roles) and differential fecundity (the selective effect of age 2 R. M. Dinkel, "Number of Children Desired by U. S. Wives and Education of Wife, Husband and her Parents for Selected Religious, Residence, and Color Groups," Proceedings of the 1969 Conference of the International Union for the Scientific Study of Population, London. 3 Definitions of high, medium, and low education differed for the two generations. 4 L. Bumpass, "Age at Marriage as a Variable in Socio-Economic Differentials in Fertility," Demography, vol. 6, no. 1 (February, 1969): 45-54.

SOCIOECONOMIC DIFFERENCES IN FERTILITY

at marriage on the relationship between education and fecundity) under­ lie the interaction. Catholic Education

In our sample of Catholic women 18-39 years of age, some 47 percent report that they had received at least part of their education in institu­ tions controlled by the Catholic Church. The association of the number iof children expected with the extent of education in Catholic schools is the same as that disclosed in other studies.5 In our data Catholic women with no education in Catholic institutions expected an average of 3.7 children; women with some Catholic education, 4.0 children; and those whose education was exclusively under Catholic auspices, 4.2 children. In a recent study8 designed to determine the basis of this relationship between Catholic education and fertility values as initially uncovered in the Princeton Fertility Study,7 the evidence assembled pointed clearly to an explanation in terms of selective factors, rather than in terms of the educational experience itself. Although limited primarily to the effect of higher education in Catholic colleges and universities, the generalization that the more religious Catholic students from homes in which Cathol­ icism is practiced are selected into Catholic educational institutions would also seem to apply to secondary schools, and perhaps to elemen­ tary schools as well. The high fertility of women educated in Catholic schools and colleges is thus to be attributed to the religiousness of the in­ dividual and her home environment rather than to the educational experience. Socioeconomic Status

We have already seen that education and expected fertility are nega­ tively associated but that the nature of the association differs somewhat for Protestants and Catholics at the college level. This difference is ex­ plained by the interrelations of education and fertility with religiousness. In the present section we are concerned with the effects of the husband's occupational class and income on expected fertility, and especially with 5C. F. Westoff, R. G. Potter, Jr., and P. C. Sagi, The Third Child (Princeton: Princeton University Press, 1963), pp. 94-102. P. K. Whelpton, A. A. Campbell, and J. E. Patterson, Fertility and Family Planning in the United States (Princeton: Princeton University Press, 1966), pp. 84-88. 6 C. F. Westofl and Raymond Potvin, College Women and Fertility Values (Princeton: Princeton University Press, 1967). 7 C. F. Westoff, R. G. Potter, Jr., P. C. Sagi, and E. G. Mishler, Family Growth in Metropolitan America (Princeton: Princeton University Press, 1961), pp. 199-

202.

CHAPTER IV

the comparative effects these variables have on the fertility expectations of Protestants and Catholics. For Protestants, there is a weak inverse relationship between the social status of the husband's occupation and the number of children expected (Table IV-16)—the type of relationship frequently observed in studies of differential fertility. Among Catholics, the association is quite differ­ ent: the fertility expectations of women married to men in the upper white-collar class and in the farm group are higher than those for other occupational categories. Catholic women with husbands in farm occupaTable IV-16. Humber of children expected by Protestant and Catholic women, by husband's occupational class, husband's income, and family income.

Husband's Occupational Class Total

Number of Children Expected

Number of Women

P

C

C-P

P

C

3.0

3.9

.88

1864

809

2.8 2.8 3.1 3.3 .3.2

4.1 3.6 3.9 3.8 4.7

1.31 .85 .79 .46 1.44

492 243 445 509 89

219 111 214 212 31

2.8 2.9 2.9 3.1 3.0 3.1 3.1 3.1 3.3

4.6 3.9 3.8 4.0 3.7 3.8 3.8 4.3 3.8

1.83 1.02 1.00 .90 .74 .72 .68 1.22 .49

75 236 235 211 271 326 208 135 120

45 117 122 99 124 133 77 38 29

2.6 2.7 3.0 3.1 3.1 3.1 3.3 3.2) 3.7/

4.2 3.8 3.8 4.0 3.9 3.7 4.2 4.3

1.57 1.09 .83 .92 .84 .58 .84 .95

107 334 333 210 230 244 144 81 68

62 156 137 107 101 94 60 22 15

Occupational Class Upper Lower Upper Lower Farm

white-collar white-collar blue-collar blue-collar

Husband's Income $15,000 10,000 8,000 7,000 6,000 5,000 4,000 3,000 0

and over - 14,999 - 9,999 - · 7,999 - 6,999 - 5,999 - 4,999 - 3,999 - 3,000

Familv Income $15,000 10,000 8,000 7,000 6,000 5,000 4,000 3,000 0

and over - 14,999 - 9,999 - 7,999 - 6,999 - 5,999 -

A,999

- 3,999 - 3,000

SOCIOECONOMIC DIFFERENCES IN FERTILITY

tions represent only a very small fraction (4 percent) of the Catholic sample. The upper white-collar category is, however, numerically im­ portant—28 percent of the Catholic total—as well as sociologically sig­ nificant (in the sense that they constitute the social elite of the Catholic population). As we observed in our discussion of the high fertility expected by Catholic college women, upper-class Catholics are dispro­ portionately drawn from the more religious segment of that population. Partly as a consequence of this difference in the composition of the upper white-collar classes, the difference between Protestant and Catholic fertility (excluding the farm category) increases with occupational prestige—the higher the class level, the greater the excess of Catholic over Protestant expected fertility. The same generalizations apply to the interrelations between religion, fertility, and income. For both the husband's income and the family in­ come—but more strongly for the latter—the average number of children expected by Protestant women decreases as income increases. The 6 per­ cent of Protestant women in families with annual incomes of $15,000 and over expect an average of 2.6 children while the 4 percent with family incomes under $3,000 expect an average of 3.7 children. On the other hand, the relation of income to expected fertility among Catholic women shows very little pattern at all, except for the high fertility of the uppermost income category. By husband's income, the highest average number of children expected by Catholics (4.6) is in the highest income bracket ($15,000 and over); by family income, the highest expected fertility appears at both the highest and the lowest categories. Again, as in the occupational class comparison, the difference between Protestant and Catholic fertility tends to be greater at the higher income levels and less at the lower levels. In general, the social-class differential in fertility is radically different for Protestants and Catholics in the United States. Whereas success, affluence, and prestige imply smaller families for Protestant couples, class has the opposite significance for Catholics. It should be noted, how­ ever, that this difference applies only at the top levels; fertility expecta­ tions for Catholics at levels just below this, for example at the lower white-collar class level, are the lowest of those at all levels. Whether this pattern of high fertility among the Catholic socioeconomic elite carries any significance for upwardly mobile Catholics depends on the nature of reference-group behavior in the American Catholic population. It is unlikely that the models for upwardly mobile Catholics are exclusively

CHAPTER IV

Catholic. It is an interesting (and researchable) question, however, whether Catholics on the way up in the socioeconomic hierarchy associ­ ate themselves with the values of the more religious Catholic elite. Wife's Employment

The lowest fertility for both Protestants and Catholics is expected by women currently working (see Table IV-17). Among working women, the career woman who works because she "likes to," rather than the woman who works primarily for "family needs" or "extra things," ex­ pects the fewest children (except among younger Catholics). The effect of working on the Protestant-Catholic differential seems slight; the dif­ ferential tends to be least among women who have never worked. Of course the same selective factors influencing the probability of a married woman's participation in the labor force that we observed in our earlier discussion of the racial differential also operate here and obfuscate the causal sequences involved. There is a considerably greater expected fer­ tility among older than among younger women in the category "never worked"; this underscores the selection of couples of high fecundity and low contraceptive effectiveness into the former group. These "facts of life" are more important than working interests for expected family size. Residence

Variations in fertility by region and size of place of residence reflect not only the cultural factors associated with urban and rural life, and differences in styles of life in various parts of the country, but also the underlying differences in socioeconomic and religious composition. (For example, income and size of place of residence are correlated; for an­ other example, Catholics tend to be concentrated in the Northeast.) In Table IV-18, we examine both the association of expected fertility with region of residence and the effect of regional residence on ProtestantCatholic differences. There is little variation in expected fertility by re­ gion of the country. For Protestant women the highest value appears in the Northeast, whereas for Catholics this region appears to produce the lowest value. Whether these differences, since they are slight, represent anything more than sampling errors is difficult to know. A similar pattern prevails with respect to variations in expected fertility by size of place of residence (Table IV-19). Aside from the dis­ tinction between rural and urban, there seems to be little significance for fertility in residence in large cities, small cities, or suburbs. For Prot­ estants, the variation in mean expected parity by size of city or suburb

3.0

3.4 3.0 2.6 2.9 2.7 2.3

1864

459 847 558 230 183 125

Never Worked Once Worked Now Working: Family needs Buy extra things Likes to work

Total

Never Worked Once Worked Now Working: Family needs Buy extra things Likes to work

P

Total

Wife's Work History and Reasons for Working

18 - 29 C-P C

3.8 3.9 3.4 3.6 3.2 3.5

3.7 .64 .86 .64 .75 .54 1.25

.76

250 398 274 124 90 52

200 382 227 79 77 64 90 181 114 37 41 33

385

Ntmiber of Women

3.2 3.0 2.7 2.8 2.7 2.3

3.0

P

209 449 284 106 93 73

942

3.6 3.0 2.7 3.0 2.7 2.2

3.1

Mean Number of Children Expected

922

.87 .92 .91 .64 .90 1.12

.88

P

809

4.3 3.9 3.5 3.6 3.6 3.4

3.9

18 - 39 C C-P

110 201 113 42 36 31

424

4.6 4.0 3.6 3.5 4.0 3.2

4.0

.98 .96 .89 .52 1.30 .94

.98

30 - 39 C-P C

Table IV-17. Number of children expected by Protestant and Catholic women* by the wife's work history since marriage and reasons for working.

CHAPTER IV Table IV-18. Number of children expected by Protestant and Catholic women, by region of residence.

Number of Women C P

Region

Number of Children Expected P C-P C

Total

3.0

3.9

.88

1864

809

Northeast Midwest South West

3.2 3.1 2.8 3.0

3.7 4.0 -4.0 4.0

.52 .88 1.13 1.03

294 592 681 297

306 255 127 121

ranges from 2.9 to 3.0, and for Catholics, from 3.7 to 3.9. For both groups, the values for rural nonfarm areas are slightly higher and for rural farm residence considerably higher. Accordingly, region and size of place of residence as controls for the study of Protestant-Catholic differences contribute no additional insights. The Trend of the Protestant-Catholic Differential

Has the religious differential in fertility diminished or increased over the decade between 1955 and 1965? This question is not as simple to answer as it might seem. If we were to focus only on the total number of children expected by women 18-39 as estimated in the three studies, we would conclude that there has been some slight expansion of the differTable IV-19. Mean number of children expected by Protestant and Catholic women, by size of place of residence.

Size of Place

Number of Children Expected P C-P C

Number of Women C P

Total

3.0

3.9

.88

1864

809

2.9

3.8

.84

91

123

2.9

3.9

1.02

255

168

2.9

3.8

.86

143

80

3.0

3.7

.72

121

56

3.0 2.9 3.1 3.5

3.8 3.9 4.0 4.9

.83 1.02 .87 1.38

469 304 315 166

137 106 103 36

Central Cities of 14 Largest Metropolitan Areas Suburbs of Largest Metropolitan Areas Central Cities 150,000 - 1,500,000 Central Cities 50,000 - 150,000 Suburbs of Cities 50,000 - 1,500,000 Smaller Cities and Towns Rural Nonfarm Rural Farm

SOCIOECONOMIC DIFFERENCES IN FERTILITY

ential: the margin of the Catholic over the Protestant mean rose from 0.5 in 1955, to 0.8 in 1960, and to 0.9 in 1965 (Table IV-20). If we restrict the question to nearly completed fertility (women aged 35-39), the ex­ pansion of the differential has been even greater—from 0.3 children in 1955, to 0.6 in 1960, and to 1.0 in 1965. On the other hand, the pattern among younger women would seem to suggest that the trend of the re­ ligious differential is reversing and heading toward contraction. The widening of the Protestant-Catholic differential between 1955 and 1960 was described by the authors of the 1960 Study as a result of an upward revision of Catholic expectations, especially for the group of women who were 18-34 in 1955 as they reached ages 23-39 in 1960. In addition, Catholics who entered the sample universe between the 1955 and 1960 Studies had higher expectations than those who left, whereas the expectations of Protestant women entering the sample were lower than those who moved out of it. Between 1960 and 1965, the expecta­ tions of Catholics leaving the sample—the 35-39 age group in 1960— were the lowest of any age category in either 1960 or 1965 and, as a birth-cohort reading of Table IV-20 reveals, expectations again rose ir­ regularly with age. This phenomenon of fertility expectations increasing with time can be seen quite clearly from a marriage-cohort reading (on the diagonals) of the second tier of Table IV-20. The increase with dura­ tion of marriage in the average number of children expected by Catholic women is greater than the corresponding increase among Protestants; the result is a widening of the religious differential over the cohort's marriage duration. But if we examine the difference between Protestant and Catholic fertility in each marriage-duration category (reading hori­ zontally) we see very little evidence of change in the differential between 1960 and 1965, although the widening between 1955 and 1960 is quite apparent. Looking at the religious differential by age at (first) marriage, we see a widening between 1955 and 1960 but little change between 1960 and 1965 for women married before age 22 (the large majority). There is some increase in the differential for women married at ages 22-24, be­ cause of a greater increase in Catholic than in Protestant expectations. The sharp increase in the religious differential among women married after age 25 (the minority) is due to a decline in the expected fertility of Protestant women, but this is probably a sampling vagary; the obser­ vation is difficult to interpret considering the age-at-marriage spread for this broad category. In conclusion, we offer the generalization that the widening of the

2.7 2.9 3.1 3.1

3.2 3.1 2.8 2.6 2.3

Number ο£ Years Married 2.8 3.5 Under 5, 2.8 3.3 5 - 9-.rf: 3.0 3.3 10-14 3.0 3.5 15+

Ase at (First) Under 18 18 - 19 20 - 21 22 - 24 25 - 39

Marriage 3.3 3.7 3.0 3.6 2.7 3.5 2.7 3.4 2.1 2.8

2.8 3.1 3.0 2.8

2.9 3.0 2.9 2.8

Aae of Wife 18-24 25 - 29 30 - 34 35 - 39

4.1 3.8 3.7 3.6 3.5

3.5 3.9 4.1 4.0

3.6 4.1 3.8 3.4

2.9 3.7

3.5 3.0 2.9 2.7 2.0

2.7 3.0 3.0 3.3

2.9 3.1 3.1 3.0

4.2 3.7 3.9 3.9 3.6

3.5 3.9 4.1 4.1

3.5 4.0 4.1 4.0

3.0 3.9

.4 .6 .8 .7 .7

.7 .5 .3 .5

.9 .4 .4 .3

.5

.9 .7 .9 1.0 1.2

.8 1.0 1.0 .9

.8 1.0 .8 .6

.8

.7 .7 1.0 1.2 1.6

.8 .9 1.1 .8

.6 .9 1.0 1.0

.9

Mean Number of Children Expected 1960 1965 C-P 1955 1960 1965 C C P P

3.8 3.4 3.3 3.1

2.9 3.4

1955 P C

Total

Characteristic

Number of Women 1960 1965 P C P C

435 531 427 296 128

406 552 476 383

394 461 505 457

92 179 178 212 126

212 278 186 111

146 220 212 209

160 189 196 123

126 176 174 192

395 93 507 182 361 177 219 154 113 62

364 417 447 367

370 396 405 425

506 621 432 221 84

419 479 451 515

500 422 472 470

138 241 232 141 57

215 208 203 183

216 169 227 197

1817 787 1596 668 1864 809

1955 P C

Table IV-20. Number of children expected by Protestant and Catholic women, by wife's age, number of years married, and age at marriage: 1955, 1960,.and 1965.

SOCIOECONOMIC DIFFERENCES IN FERTILITY Table IV-21.

Number of children expected by Protestant and Catholic women, by education of wife, for the 1955, 1960 and 1965 Studies.

1955 Education

P

C

1960 P C

1965 C P

1955

C - P 1960

1965

Ntimber of Children Expected Total

2.9

3.4

2.9

3.7

3.0

3.9

.5

.8

.9

College High School 4 High School 1-3 Grade School

2.7 2.7 2.9 3.6

3.9 3.3 3.1 3.7

2.8 2.8 3.1 3.5

4.1 3.6 3.6 4.3

2.7 2.8 3.4 3.9

3.8 3.8 3,8 4.7

1.2 .6 .2

1.3 .8 .5 .8

1.1 1.0 .4

.1

.8

Number of Women Total College High School 4 High School 1-3 Grade School

1817

787

1596

668

1863

809

306 794 457 260

73 396 208 110

284 752 392 168

79 341 168 80

393 890 425 155

130 418 173 88

Protestant-Catholic differential in fertility between 1955 and 1960 (as reflected in the average number of children expected) has not persisted between 1960 and 1965 and shows some indications of reversing. In the remaining analysis of the trend in the religious differential, we shall examine whether or not the balance of Protestant-Catholic fertility has changed over time among women of differing educational levels and among women who have completely planned their fertility. Among women who attended college, there has been no discernible trend in the Protestant-Catholic fertility differential (Table IV-21). Be­ cause of small numbers (especially in the Catholic sample), we have combined women who graduated from college with those who attended college but never graduated; this has probably obscured some changes, since, as we observed earlier, the fertility behavior of Catholic women with at least 4 years of higher education is quite different from that of Catholic women with 1 to 3 years of college. Among high school grad­ uates, there appears to be a steady widening of the religious differential resulting from a steady increase in Catholic expected family size during a decade when Protestant fertility remained stable. Among women with less than 4 years of high school education, a significant divergence oc­ curred only between 1955 and 1960. The last analysis of the trend in the religious differential focuses on the question of whether the relationship of Protestant to Catholic planned

CHAPTER IV Table IV-22.

Kumber of children expected by Protestant and Catholic women, by fertility planning status.

Fertility Planning Status

1955 P C

1960 P C

1965 C P

1955.

C - P 1960

1965

Number of Children Expected Total

2.9

3.4

2.9

3.7

3.0

3.9

.5

.8

.9

Completely Planned Excess Fertility

2.6 3.8

2.9 3.9

2.4 3.9

2.3 4.8

2.5 3.9

3.3 4.7

.3 .1

-.1 .9

.8 .8

1817

787

1596

668

1855

802

391 252

74 87

372 295

77 100

416 356

114 124

Mumber of Women

* Total Completely Planned Excess Fertility

Total includes couples in intermediate planning-status categories.

fertility is changing, or whether the changes we have observed are largely concentrated among groups of the population who have had more chil­ dren than they want (the "excess fertility" category). These are the most interesting categories theoretically and the only ones with an acceptable degree of comparability of definition across the three studies. The statis­ tics in Table IV-22 indicate clearly that there has been a considerable widening between 1960 and 1965 in the Protestant-Catholic differential among the most successful planners, because of a sharp increase in expected fertility among Catholic planners. Conversely, the increase in the religious differential among couples with "excess fertility" occurred between 1955 and 1960; that differential had not changed appreciably by 1965. Thus we can conclude that part of the divergence that occurred between Protestant and Catholic fertility between 1960 and 1965 re­ flected a differential change in the number of children desired rather than in the success of controlling fertility. THE SOCIOECONOMIC AND RESIDENTIAL DIFFERENTIALS

This chapter has focused primarily on racial and religious differences in fertility in 1965, and on the trend of these two differentials in recent years. Socioeconomic and residential classifications have been introduced principally to elucidate changes in the relation of Protestant to Catholic

SOCIOECONOMIC DIFFERENCES IN FERTILITY

fertility. Although occasional observations have been made about other differentials, no attempt has been made to record systematically the changes over time in the relation of fertility to socioeconomic status or to such conventional categories as regional and rural-urban residence. The remaining section of this chapter is devoted to the task of comparing the patterns of such associations in 1955, 1960, and 1965, with emphasis on the question of whether or not these traditional differences in fertility are widening or narrowing. The only comparable data from all three studies are for white women, 18-39 years of age. Socioeconomic Status

The educational differential in fertility seems to have widened some­ what by 1965 as a result of higher expectations among women with less than four years of high school (Table IV-23). In both 1955 and 1960 the difference between the mean number of children expected by college women and by women at the opposite end of the scale, those with no high school education, was 0.7 children; in 1965, because of higher ex­ pectations among women with the least amount of education, the range had increased to 1.2 children. Table IV-23.

Number of children expected, by wife's age and education: 1955, 1960, and 1965.

Age and Education

Number of Children Expected 1955 1960 1965

Number of Women 1960 1955 1965

18 - 39 College High School 4 High School 1-3 Grade School

2.9 2.9 3.0 3.6

3.0 3.0 3.3 3.7

3.0 3.1 3.5 4.2

417 1236 681 377

427 1153 579 255

571 1377 627 256

276 667 345 90

295 710 282 166

18 - 29 College High School 4 High School 1-3 Grade School

3.2 3.0 3.2 3.4

3.2

2.9

191

3.0

3.0

618

3.3 4.0

3.5 4.1

340 119

180 564 282 87

2.7 2.8 2.8 3.7

2.9 3.0 3.3 3.5

3.1 3.2 3.6 4.3

226 628 341 258

247 589 297 168

30 - 39 College High School 4 High School.1-3 Grade School

CHATTER IV

Part of this change reflects movement out of the sample by older women with lower education, and thus the increasing selectivity of women at this educational level. This kind of problem is intrinsic to the measurement of variables undergoing rapid secular change and makes comparisons over time very difficult to interpret. It means something quite different in 1965 not to have gone to high school than it did 10 years earlier, and, at the other end of the scale, college education has obviously become less selective, even over this short period of time. The problem of secular trends changing the distribution of the population is not confined to education: people in rural areas, and/or those with in­ comes under $3,000 are becoming more and more selective on charac­ teristics associated with fertility, such as attitudes toward contraception and efficacy of use. The income variable is further complicated by changes in the value of money as a result of inflation; for this reason we have simply not tried to make comparisons over time based on income. Some further insight into the changing association of education and fertility can be gained by examining the trends among younger and older women. Since the values of expected family size for younger women are more subject to change, they must be interpreted with more caution. Taking them at face value, the differential among younger women seems to have increased mainly because of the reduced expectations of college women, while among older women the differential widened because of the increased expectations of those with the least amount of education. In view of the rapidly diminishing numerical importance of the popula­ tion in the category with no exposure to high school and because of the continuing diffusion of knowledge of birth control techniques (especially the pill) throughout the population, the demographic significance of edu­ cational differentials in fertility will probably diminish by 1970. Al­ though we may still continue to see the highest fertility expectations at the lowest educational levels, these groups will comprise less and less of the population. The occupational differential in fertility (Table IV-24) shows less of a change between 1960 and 1965 than the educational pattern. Similar to the higher fertility expectations of the lower educational categories in 1965, the blue-collar classes show approximately 10 percent higher expected fertility values than in 1960. Generally speaking, however, the change in the pattern of fertility by occupational class has not altered appreciably. Aside from the difference in fertility between women with husbands in urban or farm occupations, there is little variation to discuss.

SOCIOECONOMIC DIFFERENCES IN FERTILITY Table IV-24. Number of children expected, by husband's occupational class: 1955, 1960, and 1965.

Husband1s Occupational Class Total Upper Lower Upper lower Farm Other

White-Collar White-Collar Blue-Collar Blue-Collar

Number of Children Expected 1955 1965 1960

Number of Women 1965 1960 1955

3.0

3.1

3.3

2713

2414

2832

2.8 2.8 2.9 3.1 3.7 3.1

3.0 3.0 3.2 3.2 3.5 3.0

3.2 3.0 3.4 3.5 3.7 2.9

620 286 644 765 242 156

725 312 465 670 154 88

779 379 685 751 125 113

Residence

The study of regional variations in fertility in the United States, except for the analysis of racial differences, has been of little interest to social science for some time. As the fertility expectations data in Table IV-25 show, there is little reason to renew that interest, because there is very little variation to discuss. Perhaps more to the point is the fact that there is probably little intrinsic cultural quality still differentiating regions in terms significant for fertility; underlying socioeconomic differences among regions are a more cogent level of analysis. And since regions of the country are becoming more similar in education and income and in­ creasingly more uniformly exposed to sets of national norms diffused through the mass media, the whole concept of region has become less important in differentiating not only fertility but behavior of all kinds. A similar argument could be developed about the analysis of fertility variations by size of place of residence. Aside from the difference beTable IV-25. Number of children expected, by region of residence: 1955, I960, and 1965.

Region

Number of Children Expected 1955 1960 1965

Number of Women 1955 1960 1965

Total

3.0

3.1

3.3

2713

2414

2832

Northeast Midwest South West

3.0 3.1 3.0 3.0

3.2 3.3 3.1 3.3

3.4 3.4 3.1 3.3

669 853 758 433

581 715 744 374

650 870 827 485

CHAPTER IV

tween the rural farm category and the rest, the characteristics relevant to fertility that once differentiated small and large cities have undoubt­ edly diminished in recent decades. The present evidence would seem to support this generalization: in none of the three studies has there been any appreciable variation in expected fertility by size of city of residence, including the smallest places classified as rural nonfarm (Table IV-26). Only the rural farm category appears to have retained its distinctiveness, Table IV-26.

Size of Place of Residence

Number of children expected, by size of place of residence: 1955, 1960, and 1965.

Number of Children Expected 1955 1960 1965

Number of Women 1955 1960 1965

Total

3.0

3.1

3.3

2713

2414

2832

Cities 50,000 or More Suburbs of These Cities Cities Less than 50,000 Rural Nonfarm Rural Farm

2.9 2.9 2.9 3.1 3.7

3,1 3.1 3.0 3.1 3.6

3.3 3.2 3.2 3.3 3.8

854 715 429 431 284

614 980 282 335 203

683 1083 428 430 208

and its demographic import for the nation has diminished appreciably because of its declining numerical importance. SUMMARY Negro women expect to have more children than white women be­ cause they already have more, but the number they say they want is, on the average, about the same as for whites. This indicates the racial difference in the control of fertility. The less education a woman re­ ceives, the more the number of children expected exceeds the number desired—a generalization that holds for both whites and Negroes. The number of children desired varies inversely with education only among Negroes, and this association is greatly diminished when parity is held constant (minimizing the effect of post factum rationalization of un­ wanted births). The racial differential in total number of children expected varies directly with the age of the woman. The same pattern holds for the num­ ber desired, except that in the youngest group (those under 25) Negroes want fewer children than whites. Age at marriage also relates to the racial differential—the older the age at marriage the smaller the number of children desired by Negroes in comparison to whites; conversely, the

SOCIOECONOMIC DIFFERENCES IN FERTILITY

younger the age at marriage the larger the number of children expected by Negroes in comparison to whites. There are particularly sharp inverse relationships between indices of socioeconomic status and the number of children expected by Negroes. The association is especially striking for family income, where the aver­ age number of children expected ranges from 2.1 for families with incomes of $15,000 or more to 5.3 for families below the poverty line. For whites the relationship follows the same pattern, but less closely. In all but the lowest (family) income classes, Negro women express desires for fewer children than white women, whereas they expect more children than whites in all but the uppermost income classes. Women who are working or who have worked expect (and, to a lesser extent, want) fewer children than those who have never worked. Those who are working because they "like to work" expect the smallest families of all. The interpretation of these relationships is complicated by differ­ ent types of selectivity. The racial differential tends to be greatest for women who have never worked and least for women currently working. The highest fertility indications for Negro women are for those living in the South, and especially for those on farms; outside of the South, the fertility expectations of Negroes correspond closely with those of whites. The expected fertility of Negroes with no farm residence in their history is quite similar to that of whites; with the ever-diminishing importance of rural life in this country, the rapid disappearance of racial differences in fertility seems imminent. However, analysis of the trend in the racial differential since 1960, based on limited information, indicates little sys­ tematic change to 1965. Turning to religious differences in fertility, we find the following: (1) Catholic women both want and expect more children than either Prot­ estant or Jewish women, a differential that is maintained regardless of the number of children they already have; (2) There is little difference in fertility among women in the various Protestant denominations; (3) Various indices of religiousness among Catholics, including both meas­ ures of formal associational involvement and community involvement, reveal the expected direct relationship with fertility; (4) For Protestant women there is a sharp negative relationship between the number of chil­ dren expected and the wife's educational attainment, whereas among Catholics a similar relationship prevails only below the college-graduate level; (5) Catholics with four or more years of higher education are a distinctive high-fertility group, and because of this the largest ProtestantCatholic fertility differential is found at the college-graduate level; (6)

CHAPTER IV

A similar Protestant-Catholic pattern prevails for occupational status and for income—the higher this level, the more the Catholic exceeds the Protestant expected fertility; (7) The lowest fertility for both Protestants and Catholics is expected by women currently working, and, among working women, the career woman expects the fewest children; (8) Re­ gion and size of place of residence do not contribute to any further understanding of the religious differential; (9) Comparison of the Protestant-Catholic differential in expected fertility in 1955, 1960, and 1965 shows no continuation between 1960 and 1965 of the widening that occurred between 1955 and 1960. (Among younger women, moreover, the differential shows some sign of contracting.) This chapter also includes an analysis of socioeconomic and resi­ dential differences in fertility and trends in these differentials. The only such differential that showed signs of widening between 1960 and 1965 was the level of the wife's education, but various considerations lead to the prediction that the demographic significance of educational differ­ entials will diminish by 1970. ADDENDUM Methodological Note on a New Measure of Number of Children Wanted

Since this book was finished, another analysis of the number of chil­ dren wanted, based on the 1965 National Fertility Study data, has been completed8; the results have both methodological and substantive sig­ nificance. As we have noted on several occasions in this chapter, we suspect that the number of children desired as conventionally measured in fertility surveys contains a considerable amount of post factum ration­ alization of children who at the time of conception had never been wanted but whose presence had been subsequently accepted and who, by the time of the interview, had been redefined and included in the total number of children desired. One suspicion of such rationalization stems from the fact that desired family size increases with age and marriage duration. Another suspicion stems from the fact, noted above, that about one-half of the women who reported an unwanted birth also reported, in a different part of the interview, that they did not prefer fewer chil­ dren than they had had. Not all of this should be regarded as inconsist­ ency since the two measures refer to different points in time—one the time of conception and the other a varying period of time later when the 8 L. Bumpass and C. F. Westoff, "The 'Perfect Contraceptive' Population," Science, Vol. 169 (September 18, 1970): 1177-1182.

SOCIOECONOMIC DIFFERENCES IN FERTILITY

woman was interviewed. For many women the additional child may now really be desired. The critical measurement for some purposes, however, is not the capacity to accommodate oneself to the unintended conse­ quences of sexual union but rather the number of children that would have been born if perfect fertility control were available, that is, the num­ ber of children the couple would have had if the only determinant of this number (with existing levels of fecundability) were the preferences of the parents at the time of conception. The approach we followed was to construct a variable that we here label the number of children "wanted." This variable is defined as fol­ lows: for women who intend to have more children, their value is the total number intended; for women who intend to have no more children, their value is their current parity minus the number reported as having been unwanted at the time of conception. Whether the birth is classified as unwanted depends on the answers to a series of questions asked about the circumstances of each pregnancy in the woman's history. If she reported not having used any contraception during the interval preced­ ing that conception but not wanting a child as soon as possible, or if she reported having used contraception but having failed to prevent a preg­ nancy which was not wanted at that time, she was then asked: "Before you became pregnant this time did you want to have a(nother) child sometime?" A negative response to this question was the basis for classifying that birth as unwanted. The estimate of the number of children wanted for all women in our sample is 2.7 children (see Table IV-27) compared with 3.3 children estimated by the conventional measure of the desired number of chil­ dren. As explained in the Bumpass and Westoff article, the demographic significance of this difference is considerable and the implications for population policy profound indeed. The prevention of unwanted births would have a pronounced effect on our rate of population growth. With respect to a policy to reduce growth, the finding supports actions de­ signed to enable people to have the number of children they want rather than attempt to change the number they want. Aside from the demographic and policy implications, the analytic use­ fulness of this approach is suggested in Table IV-27. We have suspected that the main reason for the direct association between age or marriage duration and desired family size might be not that it revealed a time trend or life cycle pattern but simply that, the longer the duration of exposure, the greater the risk of having unwanted births, some fraction of which are subsequently rationalized in statements of desired family

CHAPTER IV

Table IV-27. Number of children desired and wanted by white and Negro women, by age of wife, number Of years married, wife's education and family income.

Characteristic

All Women Desired Wanted D-W

White Desired Wanted D-W

Negro Desired Wanted D-W

Total

3.3

2.7

.6

3.3

2.7

.6

3.2

2.6

.6

Ase of Wife Under 20 20 - 24 25 - 29 30 - 34 35 - 39 40 - 44

2.8 3.1 3.2 3.4 3.4 3.5

2.5 2.7 2.8 2.9 2.7 2.7

.3 .4 .4 .5 .7 .8

2.9 3.1 3.2 3.3 3.4 3.5

2.6 2.7 2.7 2.9 2.7 2.6

.3 .4 .5 .4 .7 .9

2.6 2.8 3.3 3.5 3.2 3.6

2.4 2.5 2.8 2.8 2.2 2.7

.2 .3 .5 .7 1.0 .9

Number of Years Married TJnder 5 5-9 10 - 14 15 - 19 20 - 24 25 and over

3.0 3.2 3.4 3.4 3.5 3.8

2.5 2.7 2.8 2.8 2.7 3.2

.5 .5 .6 .6 .8 .6

3.0 3.2 3.3 3.4 3.5 3.7

2.6 2.7 2.8 2.8 2.7 3.1

.4 .5 .5 .6 .8 .6

2.7 2.9 3.6 3.3 3.7 4.1

2.3 2.5 2.8 2.5 2.5 3.5

.4 .4 .8 .8 1.2 .6

Wife's Education College 4+ College 1-3 High School 4 High School 1-3 Grade School

3.3 3.2 3.2 3.3 3.8

2.7 2.6 2.6 2.8 3.2

.6 1.0 .6 .5 .6

3.3 3.6 3.2 3.3 3.8

2.8 2.6 2.6 2.8 3.2

.5 .6 .6 .5 .6

2.8 2.7 2.9 3.2 4.0

1.9 2,4 2.5 2.6 2.9

.9 .3 .4 .6 1.1

Family Income $15,000+ 10,000-14,999 8,000- 9,999 7,000- 7,999 6,000- 6,999 5,000- 5,999 4,000- 4,999 3,000- 3,999 Under 3,000

3.4 3.2 3.3 3.3 3.2 3.3 3.4 3.5 3.8

2.6 2.6 2.8 2.7 2.7 2.8 2.8 2.8 2.9

.8 .6 .5 .6 .5 .5 .6 .7 .9

3.4 3.2 3.3 3.3 3.3 3.3 3.4 3.4 3.7

2.7 2.6 2.8 2.7 2.8 2.8 2.9 2.8 2.9

.7 .6 .5 .6 .5 .5 .5 .6 .8

2.7 2.8 3.0 2.8 3.0 3.1 '3.3 3.6 3.9

2.0 2.1 2.5 2.5 2.4 2.5 2.6 2.9 3.1

.7 .7 .5 .3 .6 .6 .7 .7 .8

size. The considerable flattening of the relationship between age or mar­ riage duration and the number of children wanted, is certainly consistent with this hypothesis. With the conventional measure of number of chil­ dren desired there is a progression with age and marriage duration; with the new measure of the number of births wanted, the relationship dis­ appears completely with age and is considerably weakened with mar­ riage duration. The new measure does not change the relative differences between the

SOCIOECONOMIC DIFFERENCES IN FERTILITY

races or between the educational or income strata of the population. Only the absolute values are altered. We suspect that there still remains considerable rationalization in the lowest educational and income classes in reporting births as wanted that were actually unwanted at the time of conception.

CHAPTER V

Contraception JUDGING from both discussion in the popular press and increasing gov­ ernmental action in the field of family planning, we are in the midst of a veritable revolution in attitudes toward a subject which until recent years was wellnigh taboo. Has this change in attitudes toward fertility control and contraception in the public sector been paralleled by change in private attitudes? Have American married women in general radically altered their perception of the desirability of controlling the number and timing of children? Have the attitudes and contraceptive practices of Catholic women in particular been affected by the extensive publicity given to the possibility of modification of the official Church position on birth control? Have American women in fact changed their attitudes toward fertility control or their practice of contraception in recent years? And, if so, have such changes been uniform throughout the population or have different age groups or socioeconomic strata changed more than others? The three studies of fertility and fertility planning in the United States conducted in 1955,1 I960,2 and 1965 offer a unique opportunity to assess such trends in both attitude and practice in recent years.3 ATTITUDES TOWARD FERTILITY CONTROL Has the attitude of American women toward fertility control become more favorable in recent years? Strictly speaking, the data from the three surveys do not permit a precise answer because of differences in the phrasing of the respective questions and in the coding of responses; nevertheless, the procedures were sufficiently similar in the 1960 and 1965 studies to encourage confidence in their comparison.4 The ques­ tions asked in 1960 were: 1 R. Freedman, P. K. Whelpton and A. A. Campbell, Family Planning, Sterility and Population Growth (New York: McGraw-Hill, 1959). 2 P. K. Whelpton, A. A. Campbell and J. E. Patterson, Fertility and Family Planning in the United States (Princeton: Princeton University Press, 1966). 8 The material in the first part of this chapter was first prepared for the Univer­ sity of Michigan Sesquicentennial Celebration in November 1967. See C. F. Westoff and Ν. B. Ryder, "Recent Trends in Attitudes Toward Fertility Control and in the Practice of Contraception in the United States" in S. J. Behrman, L. Corsa, Jr., and R. Freedman, ed., Fertility and Family Planning: A World View. (Ann Arbor: University of Michigan Press, 1969), pp. 388-412. 4 The 1955 data are not included because the question used then did not make any allowance for women who approved of the rhythm method but objected to other methods of birth control. In addition, the question was open-ended (and thus required coding). In both 1960 and 1965 separate questions were asked about the rhythm method and they were mainly precoded.

CHAPTER V

Q. 61: Many married couples do something to limit the number of pregnancies they have or to control the time when they get pregnant. In general, would you say you are against this, for it, or what? Those who replied "against"—20 percent—were then asked: Q. 61C: Some married couples use only a natural method—rhythm or safe period—to keep from getting pregnant too fre­ quently. Would you say you are against this, for it, or what? The questions asked in 1965 were: Q. 63: Many married couples do something to limit the number of pregnancies they will have. In general, would you say you are for this or against this? All respondents, regardless of their answers, were then asked: Q. 64: Some couples use a natural method—rhythm or safe period —to keep from having too many pregnancies. Would you say you are for this or against this? The main differences between the two sets of questions appear to be: (1) The primary question in 1960 included the phrase "or to control the time when they get pregnant," while the 1965 version confined the ques­ tion to the limitation of the number of pregnancies; (2) The 1965 ques­ tions reversed the ordering of the options for and against; (3) The word "only," used in the 1960 question on rhythm, was deleted in 1965. It is impossible to estimate how serious these differences are; it seems, how­ ever, that some counterbalancing is involved. The responses of white women to these questions are presented for both samples in Table V-I. In terms of the response to Questions 61 and 63, where method is unspecified, it appears that there has been an in­ crease in the proportion of women who endorse the idea of fertility con­ trol. The increase is due primarily to a substantial change in the attitude of Catholic women. If our definition of a "favorable" attitude includes endorsement of the rhythm method only (subsequent tables include both measures), the change is less pronounced and possibly even nonexistent for non-Catholic women. Among Catholics, there appears to have been some increase in favorability—from 85 percent in 1960 to 93 percent in 1965. In particular, there has been some shift among Catholics away from restricting endorsement to the rhythm method and toward approval of fertility control in general.5 5 Although the shift might simply be the result of a more permissive style of response among Catholics—the reduction from 5 to 1 percent in the "not ascer­ tained" category may be pertinent here—other data on methods of contraception

CONTRACEPTION Table V-l. Attitudes of white women from the 1960 and 1965 Studies toward fertility control, by religion.

Attitude

,a Total (%) 1965 1960

Protestant (X) 1960 1965

Catholic a) 1965 1960

For Fertility Control, Method Unspecified

80

85

91

92

52·

70

For Rhythm Method Only

13

10

5

4

33

23

Against Fertility Control 5

4

3

4

9

6

Not Ascertained

2

1

1

-

5

1

100

100

100

100

100

100

2414

2918

1596

1907

668

846

Percent Tptal Number of Women

Includes women of other religions.

Protestant women were already overwhelmingly in favor of fertility control by 1960 and reveal virtually the same pattern of response in 1965. The particular Protestant denomination seems to make little dif­ ference, with the exception of members of fundamentalist sects, whose attitudes are slightly less favorable in 1965 than in 1960 (tabular detail not presented). As noted above, the question on rhythm was asked in 1960 only of women who replied "against" to the more general question, but was asked of all women in 1965. A substantial fraction of the latter (onethird of all white women and nearly one-quarter of Catholic women) say they are "against" the rhythm method but "for" fertility control in gen­ eral. This interesting pattern may reflect a combination of anti-Catholicism for some non-Catholics and antitraditionalism for some Catho­ lics as well as for others, an attitude shaped by negative experience with the method. Cohort Analysis

Table V-2 has been prepared to permit examination of the trend in attitude toward fertility control by birth cohort. The percentages on each upper diagonal are from the 1960 Study; those on each lower diagonal actually used by Catholics support the hypothesis of a real change in attitude. These data are evaluated later in this chapter.

Percent Favorable, Method Unspecified

88 93

53 69

1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

50 74

93 92

49 70

90 92

80 85

54 65

90 89

52 62

87 89

77 82

''includes women of other religions.

Number of Women

1960 and 1965.3

87 93

94 96

91 95

94 95

97 97

87 97

87 93

Catholic

96 98

Protestant

94 97

b Total

82 89

97 95

92 93

79 92

94 96

90 95

112 189

312 424

440 641

176 176

396 433

600 649

174 242

405 482

624 762

192 206

425 478

677 740

178 241

351 534

572 841

20-24 25-29 30-34 35-39 40-44 20-24 25-29 30-34 35-39 40-44

Total Percent Favorable, Including Rhythm Method Only

Estimates for 1960 and 1965 appear on the upper and lower diagonals respectively,

78 86

81 87

80 82

20-24 25-29 30-34 35-39 40-44

1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

Cohort

Table V-2. Percent of white women in favor of fertility control, by religion, cohort, and age:

CONTRACEPTION

are from the 1965 Study. Comparisons across cohorts of the proportions of women at the same age with favorable attitudes reveal a distinct time trend in which all cohorts seem to have participated. The increase in pro­ portion of those in favor—most of which we have seen occurs in the Catholic group—does not appear to be confined to the more recent co­ horts. But if we focus on age differences within cohorts, there seems to be some tendency for the increase in favorable attitude to have been larger among the younger women, who are probably more responsive to change. Education

Analysis of the trend in attitudes toward fertility control by the wife's education (Table V-3) reveals a pattern only for Catholic women, among whom the amount of change is directly associated with the amount of education. This differential indicates a radical change in the position of the Catholic woman who has attended college. In 1960, only 39 percent of college-educated Catholic respondents were in favor of fer­ tility control, less than in any other educational category; five years later Table V-3. Percent of white women in favor of fertility control, by religion and education: 1960 and 1965.

Education

Total3 1960 1965

Protestant 1960 1965

Catholic 1960 1965

Percent Favorable, Method Unspecified College High School 4 High School 1-3 Grade School

86 83 80 68

89 87 85 69

96 93 88 79

97 93 89 79

39 55 58 46

67 73 72 52

Percent Favorable, Including Rhythm Only College High School 4 High School 1-3 Grade School

97 95 93 82

98 97 94 82

99 98 95 87

98 97 95 89

88 88 87 74

96 96 90 75

79 341 168 80

136 439 177 94

Number of Women College High School 4 High School 1-3 Grade .School

3

427 1153 579 255

585 1422 644 267

Includes women of other religions.

284 752 392 168

400 910 437 160

CHATTER V

the proportion reached 67 percent. Comparisons of those proportions which include women in favor of rhythm only reveal that the explana­ tion lies in a substantial shift of Catholic college-educated women away from an exclusive endorsement of rhythm and toward a more general en­ dorsement of fertility control. The reasons for such a dramatic change over a short period of time can only be conjectured. It seems plausible that educated women would have been more exposed to the publicity about the discussions on birth control within the councils of the Catholic Church; the attitudes of some have probably been affected by the atmosphere of uncertainty and doubt about the official position. However, even if this interpretation is correct, it does not explain why Catholic women who attended college were so little in favor of fertility control in 1960. The explanation seems to lie in the fact that Catholic women with more education tend to be more re­ ligious and are more likely to attend Catholic educational institutions. Such women tend to be disproportionately Irish in origin, and Irish Catholicism tends more toward orthodoxy. Religion of the Couple

Thus far our observations about the influence of religion on attitudes toward fertility control have been restricted to the wife's religious af­ filiation irrespective of her husband's religion. Table V-4 has been prepared to show the influence of the husband's religion and the trend in attitudes among couples in terms of the religion of both spouses. It is quite clear from this tabulation that the attitude of the Catholic-wife/ Protestant-husband combination—formerly more like that of the Catho­ lic couple than like the Protestant-wife/Catholic-husband couple—has moved sharply toward the Protestant position. In the most general terms, the fact of being Catholic is becoming less significant as a factor in shapTable V-4.

Percent of white women in favor of fertility control, by religion of wife and husband: 1960 and 1965.

Religion Wife Husband

Method Unspecified 1960 1965

Ehythm Only 1960 1965

Total Favorable 1960 1965

Number 1960 1965

Protestant

Protestant

91

92

5

4

96

96

1454

1701

Protestant

Catholic

78

90

10

6

88

96

106

127

Catholic

Protestant

59

83

28

11

87

94

114

133

Catholic

Catholic

52

67

34

25

86

92

525

691

CONTRACEPTION

ing attitudes toward fertility control and, as we shall see subsequently, in the practice of contraception itself. Religiousness of Catholic Women

Past research on social factors influencing Catholic fertility and con­ traceptive practice has consistently disclosed strong associations between the two and devoutness, as measured by religious practice; the two earlier studies both indicated similar relationships between attitudes toward fertility control and devoutness. In Table V-5 we have tabulated Table V-5.

Frequency

Percent of white Catholic women in favor of fertility control, by frequency of receiving sacraments: 1960 and 1965.

Method Unspecified 1960 1965

Never Once a Year or Less Few Times a Year Once a Month Two or Three Times a Month Once a Week or More

a

Rhythm Only 1960 1965

Total Favorable 1960 1965

Number 1965 1960

72 78 59 48

83 85 72 70

19 15 29 35

7 10 21 23

91 93 88 83

90 95 93 93

84 61 160 191

145 92 169 177

35 33

65 47

49 47

29 46

84 80

94 93

78 94

107 153

In 1960, the question referred to receiving sacraments and in 1965 to receiving Communion.

attitudes toward fertility control by the frequency with which Catholic women received the sacraments. Several features of this analysis are noteworthy. If we include in the concept of "favorable" attitude the ex­ clusive endorsement of rhythm, Catholic women are now uniformly (and nearly unanimously) in favor of fertility control regardless of their de­ gree of devoutness. On the other hand, devoutness still strongly differ­ entiates this attitude if those in favor of rhythm only are not included. The largest change in the 5-year period seems to have occurred among Catholics who are more than nominally and less than extremely devout, that is, women who report receiving sacraments more than a few times a year but less than once a week. The category of the most devout women, however, has experienced some increase in overall favorability despite the absence of change in the proportion endorsing rhythm only. Trend in Attitude among Nonwhite Women

Comparison of the change in attitudes among white and nonwhite women (Table V-6) reveals a trend toward convergence; a difference

74 76 86 83

86 83 80 68

Region Northeast Midwest South West

Education College High School 4 High School 1-3 Grade School

-3 -2 16 21

*

*

89 85 64 47

-2 -1 26

12

76 77 60

68

Base too small (N < 20)·

80

Total

W

89 87 85 69

82 83 87 91

85

85 87 SO 72

88 77 79 95

82

4 0 5 -3

-6 6 8 -4

3

Percent Favorable Method Unspecified 1960 1965 NW W-NW W NW W-NW

97 95 92 81

90 92 95

93

5 14 14

0

*

*

97 90 78 67

0 5 21

12

90 87 74

81

97 97 94 82

95 94 94 97

95

94 94 90 84

95 90 89 98

91

3 3 4 -2

0 4 5 -1

4

37 73 86 74

41 74 136 19

270

106 285 290 156

158 131 502 46

837

Total Percent Favorable Including Rhythm Method Only 1960 1965 Number of nonwhites W NW W-NW W NW WrNW 1960 1965

Tabic V-6. Percent of white and nonwhite women in favor of fertility control, by region of residence and education of wife: 1960 and 1965.

CONTRACEPTION

of 12 percent in 1960 has diminished to 3 percent by 1965. Subdivision by region of residence and wife's education indicates that changes in the South and among the less educated nonwhites are responsible for this convergence, although the small numbers of nonwhites in the samples require caution. The same pattern of change, viewed from a different perspective, is manifested in the sharp reduction among nonwhites in the degree of association between education and attitude over the five years, compared with little change among whites. A substantial part of the ex­ planation of convergence is the rapidly improving educational level of the nonwhite population as reflected in successive samples of women 18-39 years of age. Attitude and Practice

Much of our interest in the study of attitudes toward contraception resides in the assumption that attitudes relate to behavior. Ignoring ques­ tions of the temporal sequence of attitude and practice, and of their interaction, the evidence in Table V-7 clearly sustains this assumption. Among Catholic women (for whom there is sufficient variation in atti­ tude to make the question interesting) there is a rather strong associa­ tion between attitude and practice.® Table V-7.

Attitude

Attitudes of white Catholic women toward fertility control, by type of contraception used.

Never Used Any Method

For Fertility Control, Method Dnspecified For Ehythm Method Only Against Fertility Control All Catholic Women3

3

Have Used Only Rhythm

Have Used Any Other Method

Percent Total·

Number of Women

15

19

66

100

588

33

47

20

100

196

64 22

9 25

27 53

100 100

56 846

Includes 6 women whose responses to both questions do not permit classification.

6 The association is diluted of course by such factors as sterility and the number of recently married women who have not yet begun to use contraception, but will presumably do so eventually.

CHAPTER V

THE USE OF CONTRACEPTION

In the first section of this chapter, we have reported a trend over time toward a more favorable attitude toward contraception; we turn now to the question of whether or not a similar trend prevails in the extent of contraceptive use. There are differences in the details of the specific questions (used in the three studies) which form the basis for our estimates of the trend in contraceptive use. Although the wording of the main question varied somewhat in each study,7 the major difference is that both the 1955 and the 1960 Studies asked a general question about whether or not the couple have ever used any method, whereas the 1965 Study asked specific questions about use in each interval in the context of a pregnancy history, beginning prior to the first pregnancy. (Estimates of the propor­ tion of women who had ever used contraception were then derived by cumulating the experience of successive intervals.) The procedure fol­ lowed in 1965 and that used in the previous studies would appear to have opposite effects. On the one hand, the use of repeated questions for suc­ cessive intervals to derive estimates of use may have reduced underre­ porting in 1965; on the other hand, the absence of follow-up questions probing the possible use of methods not listed on the card shown to the respondent may have increased underreporting. Other procedures were similar. In all three interview schedules the questions on attitude preceded those on use. In all three studies, the rhythm method was considered as contraception. And, although details vary, a distinction has been consistently maintained between what has been called use on an "action" and use on a "motive" basis; the present analysis treats only the use of contraception for the explicit purpose of controlling the timing and/or number of pregnancies.8 7

The main questions in the three studies were:

1955—Q. 43. Now in your own case, have you or your husband ever done

anything to limit the number of your children or to keep from having them at certain times? 1960—Q. 65. Here is a card with the names of methods some married couples use to keep from getting pregnant. Have you or your husband ever used any of them? If the wife said "Yes" she was asked which methods had been used. If she said "No," she was asked: Q. 65b. Have you ever used any methods not shown on this card? 1965—Q. 100. Here is a card with the names of methods couples use to delay or prevent having a baby. During this time, which method or methods, if any, did you or your husband use? In the 1965 study, this question was repeated for each interval of exposure to risk of conception. 8 This excludes the use of douching for cleanliness only, as well as the use of

CONTRACEPTION

The Extent of Use in the United States

As of 1965, 84 percent of white married women 18-39 years of age report having used some method of contraception (Table V-8). If we add to this figure the number of nonusers to date who say they expect to Table

V-8. Percent of white couples who have used or expect to use contraception, by fecundity, duration of marriage, and parity: 1955, 1960, and 1965.

Percent Have Used 1955 1960 1965 Total

Percent Have Used or Expect to Use 1965 .1955 1960

Number of Couples 1955 1960 1965

70

81

84

79

87

90

2713

2414

2912

83 45

89 62

93 63

91 55

96 68

97 72

1794 919

1674 740

2030 882

65 75 73 65

75 86 82 78

82 87 86 82

81 83 79 68

91 91 86 80

93 93 89 84

649 869 686 509

544 649 702 519

661 755 719 778

42 71 78 81 73 67 57

55 74 89 89 87 80 76

56 81 89 91 90 90 81

59 82 84 87 78 74 65

72 85 93 92 90 84 78

75 90 92 93 94 93 84

419 603 843 468 190 104 86

301 463 682 499 263 119 87

358 491 758 613 372 161 160

Fecundity Fecund Subfecund

Duration of Marriage Under 5 5 - 9 10 - 14 15+ Parity 0 1 2 3 4 5 6+

use a method later (many of whom are women married only a short time), the proportion reaches 90 percent. And finally, if we exclude women reporting problems of subfecundity, we find that the proportion of fecund women who have ever used or expect to use contraception is 97 percent. Clearly the norm of fertility control has become universal in contemporary America. The widespread use of contraception in the United States has not just the "pill" for noncontraceptive reasons only. The primary measure of contraceptive use is simply whether or not the woman reports that she or her husband had ever used any method of fertility control in her history of exposure. This crude classifi­ cation does not measure regularity, effectiveness, or length of use.

CHAPTER V

developed in the past few years. In 1955, 70 percent reported having used contraception and by 1960 the figure had reached 81 percent. The increase to 84 percent by 1965 simply continues this trend. The same pattern of increase is also evident when the proportion expecting to use contraception in the future is included. With practice approaching uni­ versality, the rate of increase must of course diminish, since there is so little scope for further expansion. Among the (currently) subfecund, the changes in the proportions who have used contraception are in the same direction; this probably implies a growing tendency to use contraception earlier in married life. Duration of Marriage and Parity

Data on marriage duration and parity are useful to indicate the time pattern of contraceptive use. In all three studies (Table V-8) the pro­ portions who have used contraception are highest for 5-9 years of mar­ riage and 3 children. The decline in use for older marriages probably represents a combination of recall error, subfecundity due to age, and an increase in the tendency to use contraception (and use it earlier) among more recent cohorts. The same explanations are relevant for the small declines with advancing parity; furthermore, it is probably important that the women in higher parities are self-selected in the direction of nonuse. Accordingly it is of interest that in 1965 there is much less decline with advancing parity than there was in 1955 and 1960. Cohort, Age, and Religion

Table V-9 shows estimates of past or prospective use of contracep­ tion, by cohort, age, and religion. The data in the upper diagonal of each panel are from the 1955 Study, those in the middle diagonal are from 1960, and those in the lower diagonal from 1965. Comparing successive cohorts at the same age (reading down each column), there have been monotonic increases over the entire sequence for both Protestant and Catholic subsamples alike; this observation applies to the percent who have used as well as the combined percent of those who have used and those who expect to use. Variations by age within each cohort (reading across each row) are also generally positive. The bottom panels of Table V-9 show the differences over time be­ tween Protestants and Catholics in the proportion who have used, and in the proportion who have used or expect to use. It is apparent from both of these panels that, from cohort to cohort at each age, and for each cohort as age advances, the difference is becoming attenuated. The con-

76 83 88

58 65 79

18 18 9

1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

21 18 12

59 71 78

80 89 90

73 84 86

13 10 8

63 75 79

76 85 87

15 14 5

55 67 77

70 81 82

13 9

66 71

79 80

76 78

70 85 89

72 80 86

Catholic

89 93 94

82 89 89

Protestant

83 91 93

58 69 81

74 84 85

69 80 84

67 71

80 82

77 79

Protestant Minus Catholic 16 13 10 15 11 19 9 4 g 3 17 6 5 4

73 87 91

90 93 95

85 92 94

79 88 88

Totalb

Percent Have Used or Expect to Use .20-24 25-29 30-34 35-39 40-44

128 112 189

320 312 424

464 440 641

220 176 176

461 396 433

714 600 649

212 174 242

505 405 482

748 624 762

209 192 206

457 425 478

695 677 740

178 241

351 534

572 841

Number of Couples 20-24 25-29 30-34 35-39 40-44

f Estimates for 1955, 1960, and 1965 appear oft the upper» middle, and lower diagonals respectively. Includes women who are neither Catholic nor Protestant.

71 79 85

73 83 84

65 77 81

.Percent Have Used 20-24 25-29 30-34 35-39 40-44

1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

Cohort

Table V-9. Percent of white couples who have used or expect to use contraception, by cohort, age at interview, and religion: 1955, 1960, and 1965.a

CHAPTER V

vergence is particularly marked in the panel which combines those who have used with those who expect to use. The inference from this obser­ vation is that Catholics begin to use contraception at a later age than Protestants. It is probable that there has been some intercohort increase in the like­ lihood of using contraception before the end of the childbearing period. This is particularly so among Catholics, partly because not much in­ crease could be expected beyond the high values already recorded for Protestants, and especially since accruing sterility problems obviate for many the necessity of use. Nevertheless, it is likely that the major change, especially for the young, reflects a decline in the age at which contracep­ tion is first used rather than a rise in the likelihood of its ever being used. Age and Education

In 1955 there was a strong educational differential in contraceptive practice (Table V-10). This differential was sharply reduced between 1955 and 1960, and again between 1960 and 1965, except among women who report having used contraception and who have had only a grade school education. Essentially the same story is revealed once an age control is introduced, with the important qualification that the low proportion of grade school respondents reporting that they have used or expect to use is shown to be attributable to the reports of women now past 30; among women 18-29, that proportion has approached equality with the levels reported by women with more education. Furthermore, the proportion of women 18-39 with only a grade school education de­ clined from 14 percent in 1955 to 9 percent in 1965, and is less than 7 percent among those 18-29 in 1965. Thus, in one way or another, the last component of "never-users" is being erased from the population. The remaining discrepancy is apparently the tendency for young women with a grade school education to begin use later in marriage, because, if attention is confined to those who have already used contraception, the differential remains substantial. Religion and Education

The estimates in Table V-Il show that educational differentials in the two measures of contraceptive use prevail in each survey year within the Protestant and Catholic subsamples. The strongest difference by educa­ tion, regardless of religion, is that between the proportions for those with grade-school education, and the proportions for those at higher educa­ tional levels. However, a closer inspection of the table reveals that a pre-

cade School 1955 1960 1965

49 66 65

High School 1-3 1955 66 1960 78 1965 83

86

74 83

85 88 88

College 1955 1960 1965

High School 4 1955 1960 1965

70 81 84

51 74 70

62 78 84

70 78 85

86 83 87

68 78 84

46 71 77

69 75 83

78 89 88

85 88 88

73 84 86

46 67 59

71 82 78

77 84 89

88 92 89

73 83 84

51 59 61

62 76 84

69 80 83

80 87 87

65 77 81

59 72 75

76 85 88

83 90 92

88 93 94

79 87 90

67 81 89

80 92 95

88 93 94

89 95 97

84 92 94

61 82 95

82 85 89

87 94 93

90 95 94

83 91 93

54 71 68

78 86 81

83 90 93

92 94 93

79 88 88

58 65 65

65 77 86

72 81 86

82 90 91

69 80 84

377 255 267

681 579 641

1236 1153 1420

417 427 585

2713 2414 2912

43 31 54

166 145 209

271 278 375

76 59 124

556 513 762

76 56 39

174 137 144

347 286 309

115 121 158

714 600 649

112 75 80

176 138 140

348 290 375

112 121 166

748 624 762

146 93 94

165 159 148

270 299 361

114 126 137

695 677 740

Percent Have Used Percent Have Used or Expect to Use Number of Couples Total 18r24 25-29 30-34 35-39 Total 18-24 25-29 30-34 35-39 Total 18-24 25-29 30-34 35-39

Percent of white couples who'have used or expect to use contraception, by wife's age dnd education: 1955, 1960, and 1965

1955 1960 1965

Total

Education

Table V-10.

CHAPTER V Table V-ll.

Percent of white couples who have used or expect to use contraception, by wife's education and religion: 1955, 1960, and 1965.

Education

1955

Total3 1960 1965

Protestant 1955 1960 1965

Catholic 1955 1960 1965

Percent: Have Used Total College High School 4 High School 1·•3 Grade School

70

81

84

75

84

87

57

70

78

85 74 66 49

88 83 78 66

88 86 83 65

90 80 70 53

93 86 80 73

90 88 86 72

62 61 59 41

67 73 73 54

81 82 75 55

Percent Have Used or •Expect to Use Total College High School 4 High School 1-3 Grade School

79

87

90

83

90

91

67

80

87

88 83 76 59

93 90 85 72

94 92 88 75

92 88 79 63

96 92 87 77

95 92 90 79

71 71 68 49

82 83 80 64

89 90 85 72

Number of Couples Total

2713

College 417 High School 4 1236 High School 1-3 681 Grade School 377

a

2912

1817

1597

1902

7^7

668

845

427 584 1153 1420 579 641 255. 267

306 794 457 260

284 752 392 168

399 909 434 159

73 396 208 110

79 341 168 80

136 438 177 •94

2414

Includes women who are neither Catholic nor Protestant.

vious pattern of use by education had vanished in 1965. In 1955, and to a somewhat lesser extent in 1960, the Catholic-Protestant difference was much greater at the college level than at the lower educational levels. This is no longer the case. The marked shift of behavior and attitudes among college-educated Catholics between 1960 and 1965 is one of the most striking findings of this analysis. The outstanding difference remain­ ing between Protestant and Catholic use patterns, with educational level controlled, is at the grade school level. Since that differential is large only in the "have used" category, the implication is that it is a difference in the timing of contraceptive use. As noted above, there are reasons, re­ lated to the distribution of the population by age and education, to believe that this particular differential may grow smaller, and in any event become less significant as the proportion in the category declines.

CONTRACEPTION

Husband's Occupational Status

The changing influence of socioeconomic status, as reflected in the husband's occupational status, on contraceptive use is shown in Table V-12. The differences are small, but persistent through time; they seem to be weakening a little, because the groups with the lowest proportions using contraception show somewhat more increase than those with the highest proportions. In detail it appears that the farm category has now replaced the lower blue-collar class as the occupational category with the lowest proportion of use or expected use; this observation must be tempered by the consideration that the proportion of the total sample in Table V-12.

Bercent of couples who have used or expect to use contraception, by husband's occupation: 1955, 1960, 1965.

Husband's Occupation Total Upper Lower Upper Lower Farm

White-Collar White-collar Blue-Collar Blue-Collar

Percent Have Used 1955 1960 1965

Percent Have Used or Expect to Use 1955 1960 1965

Number of Couples 1955 1960 1965

70

81

84

79

87

90

2713

2414

2912

81 76 69 62 63

86 84 79 76 81

88 87 83 81 78

85 82 77 74 74

90 89 86 84 84

93 91 88 88 85

620 286 644 765 242

725 312 465 670 154

804 390 708 765 130

the farm category has shrunk from 9 percent in 1955 to 4 percent in 1965. The trend of the influence of other socioeconomic factors—income and wife's work history—has also been examined. The data are neither tabulated nor discussed in any detail here because of the difficulties of interpreting income changes over a decade, and because only the crude dichotomy of whether or not the wife has ever worked since marriage is available for comparisons over time. In brief, only a weak negative rela­ tionship between the use of contraception and the husband's annual in­ come persists; the main variation, in 1965 as earlier, is between the per­ centage of couples in the category of husbands earning under $3,000 who have ever used contraception (70 percent in 1965) and the per­ centage of those with higher incomes who have ever used contraception (82 percent in 1965). As for the division of couples on the basis of whether or not the wife has worked since marriage—an admittedly crude dichotomy—the difference is absent in 1965 as in 1955 and 1960.

CHAPTER V

Race The difference between the proportions of whites and nonwhites in 1960 who reported ever having used contraception was 22 percent for women 18-39; 5 years later the difference had diminished to 7 percent. A similar change is also evident in the proportions who have used or expect to use contraception; the racial difference has all but disappeared. The bulk of the nonwhite couples in both surveys were residents of the South, where most of the dramatic increases in contraceptive practice appear to be concentrated (Table V-13). The change in the fraction of Southern nonwhites who have ever used contraception has gone from one-half to three-quarters in the 5-year period from 1960-65, while the racial differential has declined from 32 to 12 percentage points. A similar pattern of narrowing differentials exists when the differences between the races are examined in the light of wife's education; although a positive association between the amount of education and the differen­ tial persists in 1965, it is not as strong as in 1960. In 1960 the relation­ ship between contraceptive practice and education among nonwhites was stronger than among whites; in 1965 this is no longer true. In fact, with education controlled, there are only small differences in contraceptive practice between white and nonwhite couples. Current Risk Although whether or not a couple has ever used contraception is an important attribute in its own right, it lacks many desirable characteris­ tics of good measurement. For example, it does not reflect a couple's current status and, perhaps more fundamental, it does not measure exclusively the probability of use by couples who feel they need it. For example, some couples will have discovered subfecundity; had this con­ dition not developed they would probably have used some method of contraception and accordingly been classified differently. These are just some of the problems with this measure, but they can be remedied. In Tables V-14/V-16 we present a tabulation based on a more refined measure of contraceptive use. Since this analysis is confined to data from the 1965 Study, we revert to comparisons of categories of white and Negro, and white non-Catholic and Catholic women under 45.9 The new measure is defined as the percent of couples currently at risk of concep­ tion who are not currently using contraception. The concept "at risk" means either that the couple is using a method of contraception (and is 9 Part of this material was prepared for the 1969 Roundtable of the Milbank Memorial Fund, "Contraceptive Practice Among Urban Blacks in the United States, 1965."

88 83 78 66

77 82 83 80

81

2 16 22 24

*

*

86 67 56 42

1 23 32

22

76 59 51

59

Base too small (N < 20).

Education College High School 4 High School 1-3 Grade School

Northeast Midwest South West

Region

Total

W

88 86 83 65

84 84 87 83

84

85 83 79 58

84 74 75 83

77

3 3 4 7

0 10 12 0

7

Percent Have Used 1960 1965 Ntf W-NW W NW W-NW

93 90 85 72

85 88 88 89

87

-2 9 6 15

*

*

95 81 79 57

-10 12 20

11

95 76 68

76

94 92 88 75

89 91 88 92

90

88 91 87 71

91 79 85 93

86

6 1 1 4

-2 12 3 -1

4

Percent Have Used or Expect.to Use I960 1965 W NW W-NW W NW W-NW

37 73. 86 74

41 74 136 19

270

106 285 290 156

158 131 502 46

837

Number of Nonwhites 1960 1965

Table y-13. Percent of white and nonwhite couples who have used or who expect to use contraception, by region of residence and education: 1960 and 1965.

CHAPTER V

therefore presumably at risk) or that no method is being used despite the fact that the wife is not pregnant, trying to become pregnant, classified as postpartum, or subfecund. Those who are pregnant, trying to become pregnant, in the postpartum, or subfecund are couples who are at least temporarily out of the market for contraception. The numerator of our new measure consists of couples not using contraception who are simply taking chances, who are in varying degrees indifferent or casual about the possibility of a pregnancy, or who are opposed to or ignorant of con­ traception. (We are probably somewhat overestimating the extent of such noncontraceptive practice because there is undoubtedly some under­ reporting of subfecundity in these data.) The denominator consists of this same group plus those using some method of contraception. A total of 6 percent of the couples in our sample who are at risk of conception are currently not using any form of contraception. Negro couples show a much higher percentage currently at risk but not using (13 percent) than white couples (5 percent); Catholics reveal almost the same relative distance from non-Catholics (8 percent and 4 percent respectively). As expected, the percent at risk not using is highest among the youngest women and the most recently married (Table V-14); it decreases with age and to a lesser extent with marriage duration, except for Negroes, among whom the difference between cohorts is striking. The largest racial differential is, accordingly, found mainly among the oldest women. There is also a fairly large racial difference among the youngest women (those under 20 years of age); this probably reflects simply a greater cautiousness and tendency to plan among whites. Quite a dif­ ferent pattern emerges for the religious differential, which shows no sys­ tematic patterning by age or marriage duration. The association with parity discloses that the greatest nonuse among those at risk is at the lower and at the higher values, probably indicating the presence of a casual attitude before childbearing begins in earnest and different situations at the high parity levels. For white non-Catholics and Negroes at the higher end of the scale, the high proportions of nonusers at risk are a primary reason for the high parities achieved uninten­ tionally. For white Catholics, on the other hand, while the proportion at risk not using shows some increase for women with 5 and 6 or more live births, the proportion with 6 or more is only half that for white nonCatholic women at that end of the parity distribution. This seemingly curious reversal probably only reflects the tendency for large families to be planned more by Catholics than non-Catholics. There is little regional variation in current use (Table V-15) among

7 8 3 4 3 4 12

Parity 0 1 2 3 4 5 6+

White = W; Negro = N.

21 14 10 11 3 16 19

10 12 8 19 20

21 8 11 9 17 21

13

14 5 7 7 0 12 •7

3 7 5 15 15

9 2 7 5 12 17

8

4 8 2 3 2 1 15

5 3 3 3 4

10 5 2 3 3 3

4

0

Includes non-Negro nonwhites.

18 9 9 5 4 7 8

12 9 3 6 8

17 8 10 6 9 4

8

Percent at Risk Not Using WC N WNC N-W

^ White non-Catholic = WNC; White Catholic = WC.

a

7 5 3 4 5

Duration of Marriage 8 Under 5 6 5 - 9 10-14 4 15-19 5 20+ 5

8 9 4 4 3 6 14

11 6 4 4 5 3

12 6 5 4 6 5

Age of Wife Under 20 20-24 25-29 30-34 35-39 40-44

5

6

W

Total

Totalc

14 1 8 2 3 6 -7

7 6 1 3 4

6 3 8 3 6 1

4

WC-WNC

197 415 800 652 373 172 222

587 653 593 598 399

133 542 548 586 529 493

2,831

Totalc

157 333 682 535 300 131 125

430 512 478 513 329

96 401 423 475 437 431

38 79 110 109 70 38 93

148 137 108 79 65

34 135 117 107 86 58

537

124 257 531 383 189 75 66

310 370 336 351 257

78 291 317 334 298 307

1,625

Number of Women N WNC

2,263

W

33 76 151 152 111 56 59

120 142 142 162 72

18 110 106 141 139 124

638

WC

Table V-14. Percent of couples at risk of conception not currently using contraception, by age of wife, duration of marriage, and parity, by race" and religion.

a

5 5 7 4

Base too small (N.< 20).

8 *

7 21

13

*

12

29

5

8

5

8

Includes non-Negro nonwhites.

7

11

2 4 10 14

376 108

* *

* *

4 8

66

216

99 100

9

8 15

6

4

503

564 726 620 353

35

157

0

47

298

114 76 318 29

537

N

63

209

79

385

888

305 518 540 262

1,625

W-NC

Number of Women

2,263

W

558

680 808 952 391

2,831

Total3

4

4

3 4 6 6

4

WC-WNC

1,378

7

7 8 10 8

'8

WC

1,688

3

4 4 4 2

4

8

13

7 9 15 17

WNC

N-U

N

Percent at Risk Not Using

4

5 5 5 3

5

6

Never lived on a farm 5 Once lived on nonsouthern farm 5 Now lives on nonsouthern farm 8 Once lived on a southern farm 7 Now lives on a southern farm 13

Farm Residence

Northeast Midwest South West

Region

Total

W

Totala

3

7

20

118

490

259 208 80 91

638·

WC

Table V-15. Percent of couples at risk of conception not currently using contraception, by region and farm residence, by race and religion.

CONTRACEPTION

whites. Negroes in the South and West (the small subsample in the West requires caution) show the highest incidence of nonuse while at risk. Couples currently living on a farm use contraception less, especially Negro couples living on Southern farms (where the incidence of current nonuse while at risk reaches 29 percent). The pattern of association between this measure of current contracep­ tive use and wife's education and husband's occupational status, and the differential by race and religion are essentially the same as those for the measure of total contraceptive use: nonuse decreases with increasing education and (less systematically) with occupational status (Table V-16). Nonuse of contraception is also related to income but only at the lower levels. In both races, and among non-Catholics as well as Catho­ lics, the woman who has never worked uses contraception least. Among women who work because they "like to" only 2 percent are at risk but currently not using contraception. (The complexities of causal interpre­ tation are reviewed in Chapter IV.) TRENDS IN CONTRACEPTIVE PRACTICE10

The 1965 National Fertility Study also provides the opportunity to examine changes over a decade in the methods of fertility control em­ ployed by married couples in the United States. Three circumstances make such an examination especially interesting at this time. First, the birth rate has been declining since 1957; this change cannot be explained without analysis of fertility regulation. In the second place, the oral con­ traceptive has recently emerged as the leading method of birth control. (This phenomenon is described at length in Chapter VI.) Thirdly, the Catholic Church had been actively reviewing its position on the subject of birth control during this period. Comparability of Data

In some respects the data described here are not completely com­ parable. As noted earlier in this chapter, the effect of the 1965 procedure for determining the methods of fertility control used, the repetition of questions for each interpregnancy interval, may have been to overesti­ mate the incidence of methods used at the expense of comparability with 10 These data were first described in C. F. Westoff and Ν. B. Ryder, "Methods of Fertility Control Used in the United States: 1955-1965," in William T. Liu, ed., Family and Fertility, Notre Dame, Indiana: University of Notre Dame Press, 1967, pp. 157-70. (Also published by the Population Council as "The United States: Methods of Fertility Control, 1955, 1960 and 1965," Studies in Family Planning, no. 17 [February, 1967].)

5

6

2 3 4 8 13

3 5 6 8 10

2 5 3 3 5 6 6 10 14

9 5 4 5 5 2

Total

Education of Wife College 4+ College 1-3 High School 4 High School.1-3 Grade School

Occupation of Husband Upper White-Collar Lower White-Collar Upper Blue-Collar Lower Blue-Collar Farm

Income of Husband $15,000+ 10,000-14,999 8,000-9,999 7,000-7,999 6,000-6,999 5,000-5,999 4,000-4,999 3,000-3,999 0-3,000

Wife's Work History Never worked Once Worked Now Working: For family needs For extra things Likes to work

* Base too small (N < 20). a Includes non-Negro nonwhites.

7 4 3 4 4 2

2 5 3 3 5 5 6 8 12

3 4 5 6 7

1 3 4 6 12

W

Totala

12 6 6 4 10 2

16 -3 4 6 1· 9 7

19 0 9 11 7 17 19

19 11 10 9 13 4

* *

8 2 5 9 22

*

*

11 6 10 15 29

6 4 3 10 9

8

13

7 7 7 .17 20

N-W

N

6 3 3 4 2 2

0 3 1 2 5 4 3 7 12

1 4 5 5 5

1 3 3 6 9

4

WNC

11 8 5 5 6 4

5 4 2 0 4 3

6 7 6 2 0 2 13 5 4

*

*

6 10 7 4 4 6 17 12 16

6 1 2 4

2 0 5 2 9

7 5 7 9

3 3 7 8 17

133 164 123 89 24

269 223 171 395 313 198

5 13 21 26 58 110 122 94 99

45 65 108 273 24

27 44 202 191 83

552 1,022

113 360 337 267 347 354 222 129 122

694 295 553 556 94

199 290 1,102 478 181

537

N

205 157 123

394 724

78 258 227 187 237 257 170 104 103

508 215 386 385 75

167 229 771 329 124

1,625

W-NC

Number of Women

2,263

W

697 1,198

119 374 363 294 407 469 349 228 228

745 361 668 843 119

227 340 1,310 678 274

2,831

8 4

WC-WNC Totala

WC

Percent at Risk Hot Using

64 66 48

158 298

35 102 110 80 110 97 52 24 20

186 80 167 171 19

32 61 331 149 57

638.

WC

Tabla "V-16. Percent of couples at risk of conception not currently using contraception, by education of wife, occupation of husband, husband's income, and wife's work history, by race and religion.

CONTRACEPTION

the earlier surveys. Another particular difference was a modification of the way in which information was obtained concerning the practice of douching for contraceptive purposes. A third difference was the ordering of options on the card to which the respondent was referred in replying to the question about methods used; it is suspected that a change in order is responsible for the substantial report of abstinence as a method in 1960 but not in 1955 or 1965. Methods Ever Used It is important in interpreting some of our results to remember our earlier observations that the proportion of couples reporting they had ever used any technique of contraception11 rose between 1960 and 1965, continuing a trend observed between 1955 and 1960. Between 1955 and 1960, the number of methods per couple using contraception increased by 10 percent (from 1.70 to 1.86); two methods showed significant increases—condom and jelly alone. By 1965, the number of methods per couple using contraception had increased to 1.98, perhaps in part because of the procedural change noted, but almost certainly because of the availability of a new method, the oral contraceptive. It is apparent from Table V-17 that there has been a major change in the use of various methods between 1960 and 1965. For whites and nonwhites alike, the use of rhythm, diaphragm, and condom decreased sharply in favor of the oral contraceptive. The pill was first licensed for contraceptive prescription in June 1960, and its use has increased at an accelerating rate; by the time of our interviews in late 1965, 33 percent of white women and 29 percent of nonwhite women who had ever used any method reported having used the pill. It seems that the pill has been adopted primarily by couples who would otherwise have used, or were formerly using, the diaphragm, the condom, or the rhythm method. Al­ though the douche shows an apparent increase in use, the change is probably artifactual.12 11 Couples are classified as using contraception on a motive basis rather than on an action basis. This procedure excludes behavior which is contraceptive in effect but not in stated intent, such as use of the pill for medical reasons only. The extent of the latter activity is not insignificant, particularly for Catholics. To the extent that there is dissimulation about reasons for using the pill, the estimates of the use of oral contraception reported in the present monograph are under­ stated. 12 Douching may be used for personal hygiene as well as for contraception. The reported figure refers only to women who asserted contraceptive intent. Further-

Number of Users Pill Diaphragm Condom Jelly Alone Withdrawal Ehythm Douche All Other Total

466 -

12 28 4 17 67 17 12 157

453 -•

17 25 2 13 65 18 4 144

655 25 11 28 4 15 59 20 8 169

White Catholic 1955 1960 1965

56' 72 2 6 8 11 3 158

-

64 51 74 8 4 9 8 2 156

-

101

55 25 45 67 2 9 13 14 2 177

Jewish 1955 1960 1965

^ The total exceeds 100 percent because many couples reported two or more methods used.

The 1955 sample was confined to the white population.

1362 1347 1648 36 46 41 31 56 45 48 8 14 11 18 15 17 25 27 21 28 32 32 12 12 17 183 201 209

1901 1948 2445 33 36 38 26 43 50 42 6 9 11 15 17 15 34 35 31 28 24 28 8 11 14 170 186 198

Method

a

Protestant 1955 1960 1965

Total White 1955 1960 1965

30 58 19 21 18 50 19 215

-

160

1960

651 29 17 43 19 14 11 53 35 219

1965

Nonwhitea

Table V-17. Percent of users who have ever used the specified method of contraception, by race and religion of white wives 18-39: 1955, I960', and 1965.

CONTRACEPTION

The residual category of "all other" contraceptive methods consists principally of suppositories, foam, and the intrauterine device. The last of these, which did not become generally available until 1964, is reported as a method used by 1.3 percent of white users and by 2.8 percent of nonwhite users. These proportions may be expected to increase consid­ erably in the years ahead. Suppositories were employed by 6 percent of white users and by 16 percent of nonwhite users in 1960; the propor­ tions were approximately the same (5 percent and 17 percent respec­ tively) in 1965. Use of vaginal foam was insignificant in 1960, but by 1965 the method had been employed by 7 percent of white users and by 15 percent of nonwhite users. There are some important differences in the patterns of change in the use of different methods among the major religious groupings of the white population. The methods most used by Protestant couples in 1955 and in 1960 were the condom and the diaphragm; the use of both had declined substantially by 1965. Reliance on the condom and the dia­ phragm was even more pronounced among Jewish couples in 1960; but again use of both had declined by 1965. Condom and diaphragm use declined less among Jews than among Protestants, and use of the pill is less likely for Jews than for Protestants. Among Catholic women the pic­ ture of change is dominated by the appearance of the pill between 1960 and 1965. The number of methods used per couple increased from 1.57 to 1.69 and there was a decrease in the proportion reporting that they had ever used the rhythm method. "Abstinence" was reported as a method by 9 percent of women using contraception in 1960, and by less than 1 percent in 1965; the difference is probably attributable to the change in salience of the method in the questionnaire format. Changes in Methods Most Recently Used

In Table V-17, the methods described are all those which the respond­ ents have ever used. Consequently, the experience reported may, for some of the women, extend back over 25 years prior to the date of inter­ view. In order to show the most recent behavior of the samples, Table V-18 has been prepared. It differs from the preceding analysis in that it describes the methods used most recently (in the first half of the 1960's) and compares them with the techniques of fertility control employed by American couples during the first half of the 1950's. A further difference more, douching is frequently a supplementary technique to other methods of con­ traception. Much of the use reported in this chapter involves techniques employed in combination.

Includes white women of other religions.

29 30 5 7 12 9 8 8 54 4 7

-

12 15

100 20 4 15 1 7 36 4 13

655

-

6 -

Includes 3 percent reporting use of pill in combination with some other method.

^ Includes other methods used singly and all multiple usage involving both alternate and combined use.

c

-

-

-

24 27 44

100

55

-

37 56 2 3 2

-

100

64

1965

Jewish a)

1955

^ Includes a small number of women who reported using contraception but for whom specific method is unknown.

a

25 27 4 7 22 8 7

-

100

100 30 12 19 3 5 4 7 20

-

100

100 27 10 18 2 5 13 6 19

-

100

Percent Total Pill0 Diaphragm Condom Jelly Alone Withdrawal Rhythm Douche j All Other

453

1648

1362

2445

1901

Number of users'3

1965

Catholic σ>

1965

1955

1965

1955

1955

Protestant (%)

Total Whitea(%)

Method

TatXe V-X8. Methods of contraception used most recently, by religion of white wives 18-39: 1955 and 1965.

CONTRACEPTION

is that in Table V-17 many couples are included more than once, where­ as in Table V-18 each couple is represented only once. The data for 1960 are omitted because the procedures followed in 1960 do not permit comparability.13 While the bases for the 1955 and 1965 estimates are not exactly the same,14 they are more alike than either is like the 1960 figures. The impact of the pill on the distribution of methods used is revealed more clearly in the comparison of methods most recently used than in the comparison of methods ever used. Reliance on the condom, the dia­ phragm, and rhythm—which, in terms of separate use, amounted to 74 percent in 1955—had declined to 41 percent by 1965. The pill, which is responsible for this change, had by 1965 become the most popular method of contraception used by American couples, a fact that would be even more pronounced if the comparison were restricted to the younger women in the sample. The method showing the greatest con­ comitant decline is the diaphragm, followed by the rhythm method and the condom in that order. The patterns of change are not the same for couples of different re­ ligions. Among couples where the wife is Protestant, use of the dia­ phragm and condom has declined appreciably, use of the rhythm method has virtually disappeared, and the pill now clearly dominates the picture as the most popular method. The pattern of change for Jewish couples seems reasonable, despite the small numbers in the two studies. As with Protestants, both the condom and diaphragm have declined in popularity with the adoption of the pill, but both methods continue to be used much more extensively by Jewish couples than by others. Thus 71 percent of Jewish couples still depend on these two mechanical methods, whereas only 31 percent of Protestants and 19 percent of Catholics rely on them. Among couples where the wife is Catholic, the major change has been a decline in reliance on the rhythm method from 54 to 36 percent. The extent of use of the condom has not changed, but use of the diaphragm has declined from 12 to 4 percent. The pill appears to have become the second most popular method of fertility control among Catholic couples. Recent estimates for 1969 indicate that the pill may now exceed the rhythm method in popularity (see Chapter VIII). 13 In 1960 respondents who were alternating use among two or more methods were not queried about which method they had used most recently. 14 In the 1965 Study, unlike that of 1955, a respondent who last used con­ traception in an interval prior to her most recent pregnancy and reported using methods alternately was not asked which method was used most recently. As a result, a higher proportion of couples interviewed in 1965 are classified in the category "All Other," which includes multiple method usage.

CHAPTER V

Types of Current Exposure

Women differ widely in the probability that they will conceive at any given time. Some are sterile as a result of surgery, and others are subfecund for one reason or another; some women are pregnant, others are trying to become pregnant, and still others have recently given birth. The remaining category of women not using contraception are those we dis­ cussed in our measure of current risk—those who, although presumably fecund and not deliberately trying to become pregnant, are not using contraception. All the others are currently using various methods of con­ traception; for them the probability of unintentional conception varies with the use-effectiveness of the method. There is considerable interest in the question of how couples of reproductive age contend with the risk of conception. In Table V-19 we present the distributions of types of current exposure to conception. The two basic types of noncontraceptive and contraceptive exposure are di­ vided further into subtypes. Women not currently using contraception are classified as subfecund (accounting for the large majority of such exposure among all women) and those fecund women who are either simply not using a method, trying to get pregnant, already pregnant, or in the postpartum period. Couples using contraception are classified ac­ cording to the method on which they are currently relying. The sample is further classified by age of wife, by race, and by religion. Focusing first on white-Negro differences we can see that the higher proportion of Negro couples not using contraception is due primarily to the greater tendency among Negroes not to use contraception in spite of being fecund. Evidently Negroes are more apt to use contraception casually and to interrupt its use more often without explicit fertility ob­ jectives. Older Negro women show a distinctly higher proportion of subfecundity. Turning to religious differences among white couples we see that the main sources of lower levels of Catholic contraceptive practice are: preg­ nancy, the attempt to become pregnant, or, again, simple nonuse in spite of being fecund. (The last difference holds for both younger and older women.) The proportion of subfecund CathoUc women is actually lower than that of subfecund non-Catholic women. Although part of this re­ sults from a greater tendency for non-Catholics to report sterilization (for contraceptive reasons), Catholics, because they use contraception less than non-Catholics, are more likely to provide evidence of sub­ fecundity. These forces should counterbalance one another.

Pill IUD Diaphragm Condom Foam J e l l y Al'one Suppository Withdrawal Rhythm Douche All Others

Contraceptive

Total

15 1 6 13 2 1 1 4 7 3 2

16 1 6 14 2 1 1 4 7 3 2

57

2 8 1

3 8 1

55

28 3

43

Noncontraceptive Total 45

29 3

4800 100

N u m b e r of C o u p l e s Percent Total

Subfecund Fecund nonuser Fecund, trying to get pregnant Currently pregnant Postpartum

3762 100

'r o t a l

59

18 1 8 15 3 1 1 4 3 3 2

12 2 3 9 3 2 2 3 1 10 1

2 6 1

30 2

41

2670 100

49

3 8 3

30 7

51

969 100

U n d e r 4 5 (%) N W WNC

10 1 2 11 1 1 1 6 18 2 2

54

4 11 2

25 4

46

1092 100

wc

26 1 4 12 3 1 1 2 5 3 2

60

5 15 3

13 4

40

1914 100

Total

19 3 2 10 5 2 2 3 2 9 1

58

60 27 1 4 12 3 1 1 2 5 2 2

4 13 6

30 1 5 13 3 1 1 3 1 2 2

63

4 14 2

14 3

37

42 11 7

1039 100

440 100

5 15 2

13 4

40

1444 100

U n d e r 30 G ) N .W WNC

54

52

2 14 2 2

-

1 8 2 1

-

8 1 8 15 2 1 1 5 8 4 2

1 3 1

8 20 3

19

39 3

46

2886 100

Total

11 6

48

405 100

WC

8 1 8 16 2 1 1 6 9 3 2

56

1 3 1

38 2

44

2318 100

7 1 3 8 3 2 2 3 1 10 1

40

3 4 1

45 7

60

529 100

30 - 4 4 W N

10 1 10 17 2 1 1 4 4 3 2

56

-

1 2

39 2

44

1631 100

WNC

(%)

T y p e o f c u r r e n t e x p o s u r e t o c o n c e p t i o n , f o r w h i t e s a n d N e g r o e s , for' w h i t e n o n C a t h o l i c s a n d w h i t e Catholics, by age.

Current Exposure

Table V-19.

5 1 3 12 1 1 1 8 21 2 2

55

1 5 1

34 4

45

687 100

WC

CHAPTER V

Methods Currently Used

Our discussion of group differences in methods of contraception cur­ rently used is based on Table V-20 where the basis of the calculation is the number of couples currently using any method. As disclosed in the tabulation of methods most recently used, the pill is currently the most popular. This is true for whites and Negroes as a whole and for white non-Catholics, but among white Catholics the rhythm method still domi­ nated the picture as of 1965. If, however, we confine attention to women under 30, the pill is the overwhelmingly popular choice, even among Catholics. Among all groups the condom remains popular; the douche is used primarily by Negroes, especially older Negro women. The pill and the condom collectively account for more than half of the use by white non-Catholics; rhythm, condom, and pill account for nearly threequarters of the Catholic use; and the pill, douche, and condom are used collectively by two-thirds of the Negroes. KNOWLEDGE OF THE OVULATORY CYCLE

The amount of knowledge of the biological facts of conception—the time during the ovulatory cycle when a woman is most likely to become pregnant—can be considered as an index of the reproductive sophistica­ tion of a population. In the city of New Orleans in 1965, Beasley, Harter, and Fischer found that only 13 percent of a sample of Negro women under age 45 who had at some time been married and pregnant pos­ sessed "essential knowledge" on the subject. In their study, the fertile period was liberally defined as the middle seven days of the cycle. The 1965 National Fertility Study included a question which both per­ mits some comparison with the New Orleans study and allows us to compare the knowledge possessed by Negroes and whites, and by Catho­ lics and non-Catholics. All women in our study were asked: If a woman has her period every 28 days, on which day does she have the greatest chance of becoming pregnant, counting from the first day her period begins? We have defined as correct any response from 12 to 16 days inclusive and any range answers which overlap this interval. According to this definition, less than one-quarter of the Negro women in our sample are correctly informed compared with one-half of the white women (Table V-21). The figure is higher for Negroes in our national sample than for those in the New Orleans study, despite our stricter definition.

2681 100 27 1 11 24 4 2 2 7 12 6 3

Includes non-Negroes and

Number of Users Percent Total Pill IUD Diaphragm Condom Foam Jelly Alone Suppository Withdrawal Rhythm Douche All Others

*

2154 100 27 1 11 25 4 2 1 7 13 5 3

470 100 26 4 6 19 7 4 4 6 3 20 2

Under 45 N W

nonwhites.

Total

1565 100 30 1 14 26 5 2 2 6 5 5 3

WNC

(%) 589 100 19 1 4 20 2 1 1 11 34 4 3

wc 1154 100 43 2 6 20 5 2 2 4 8 5 2

Total 867 100 45 1 6 20 5 2 1 4 8 4 3

256 100 33 5 3 18 8 3 3 4 3 17 2

656 100 47 2 8 21 5 2 1 4 2 4 4

Under 30(%) -W N WNC 211 100 37 1 2 16 3 1 1 3 27 5 3

WC 1527 100 15 1 14 27 3 2 2 10 15 7 3

Total

*

Methods of contraception c u r r e n t l y u s e d b y w h i t e s a n d N e g r o e s , w h i t e n o n - C a t b o l i c s a n d Catholics, by.age.

Method Currently Used

Table V-20.

1287 100 15 1 15 28 3 2 1 10 16 5 3

214 100 16 1 8 20 6 6 6 7 3 25 1

30 - 4 4 W N

white

909 100 18 1 18 30 4 3 1 8 7 6 3

WNC

(%)

378 100 9 1 6 22 1 1 1 15 38 3 3

WC

CHAPTER V Table V-21.

Percent of women "correctly informed" about the timing of the fertile period in the ovulatory cycle, by education, race, and religion of wife.

Education

Totala

W

N

WNG

WC

Percent Correctly Informed Total High School 4+ < High School 4

41 58 27

50 6130

Total High School 4+ < High School 4

4809 2964 1844

3770 2524 1246

22 30 15

49 59 29

53 66 30

2675 1803 872

1095 721 374

Number of Women

a

969 413 555

Includes non-Negroes and nonwhites.

There is very little difference in knowledge between Catholics and non-Catholics; if anything, Catholic women (perhaps because of their greater familiarity with the requirements of the rhythm method), seem to be somewhat better informed. The chief factor determining knowledge of this kind seems to be amount of education; there is a very large dif­ ference (about 2 to 1) between the proportion of those who at least completed high school and the proportion of those with less education who are correctly informed. This educational distinction does not, how­ ever, erase the racial difference in knowledge; Negroes are considerably less informed than whites in both educational classes. Much of this dif­ ference may be because the average number of years of schooling for Negroes is far lower than that for whites within these ostensibly homo­ geneous educational classes; nonetheless, the information on this subject possessed by Negroes with four years of high school education or more is no greater than that for whites with less than four years of high school education. STERILIZATION FOR CONTRACEPTIVE REASONS

The most extreme mode of fertility prevention is resort to surgical sterilization, a practice that, although still infrequent in the United States, appears to be increasing somewhat.15 In 1965, 14 percent of the wives aged 18-39 reported that either they or their husbands had been steri15 The remainder of this chapter is derived from C. F. Westoff and Ν. B. Ryder, "Family Limitation in the United States," a paper prepared for the meetings of the International Union for the Scientific Study of Population, London, 1969.

CONTRACEPTION

lized, including 8 percent who indicated that the intent of the operation was at least partly contraceptive and 6 percent who reported otherwise.16 Contraceptive sterilizations were almost all tubal ligations (for the female) or vasectomies (for the male); noncontraceptive sterilizations were predominantly hysterectomies. The present report is restricted to contraceptive sterilization. Incidence, Trend, and Characteristics

In 1965, the proportion of white women 18-39 who reported that either they or their husbands had undergone a sterilizing operation for contraceptive reasons was 7.4 percent, an apparent rise from 5.6 percent in 1960 and an estimated17 4.0 percent in 1955. Female operations in­ creased from 3 to 4 to 5 percent and male operations from 1 to 2 to 3 percent. The data from the three studies on contraceptive sterilization among white couples by wife's age and birth cohort (Table V-22) clearly sug­ gest an intercohort trend. The tendency for operations to be more preva­ lent at older ages reflects the fact that contraceptive control becomes a more critical problem in the later stages of childbearing both for general Table V-22.

Percent of white couples with contraceptive operations, by wife's age and birth cohort: 1955, 1960, and 1965.a

Af;e of Wife Cohort 1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

a

20-24

1 2 2

25-29

4 4 7

30-34

5 7 10

35-39 5 9 11

40-44 7 10

Estimates for 1955, 1960, and 1965 appear on the upper, middle, and lower diagonals respectively.

16The relevant series of questions in 1965 was: Q. 53 "Have you or your (present) husband had an operation that makes it impossible to have any more children?" If yes: Q. 54 "What kind of operation was it?" (A list of the most common operations was provided and the date ascertained.) Q. 56 "Was the op­ eration done at least partly so that you would not have any more children?" The question asked in the 1960 Study was not the same. The determination of whether an operation was contraceptive or remedial in intent was made on the basis of an open-ended question: Q. 37a "Why did (you, he) have the operation?" 17 The 1955 Growth of American Families Study did not collect information necessary to classify sterilizing operations as contraceptive or not. The estimates for 1955 were made by the investigators in the 1960 study.

CHAPTER V

as well as for health reasons and that sterilization is a terminal solution. This interpretation is further supported by data from both the 1960 and 1965 Studies (Table V-23) showing a sharp tendency for contraceptive operations to increase with the number of births and to be highly con­ centrated among couples in the excess fertility category (couples who reported that at least their last child was unwanted). Table V-23.

Percent of white couples, wife 18-39, with contraceptive operations, by number of births and fertility-planning success: 1960 and 1965.

Percent with Contraceptive Operations 1960 1965 Total

Number of Couples 1960 1965

6

7

2414

2919

1 1 4 7 14 14

0 0 4 10 15 19

301 463 682 499 263 206

361 492 758 615 372 321

3 15

3 17

515 413

596 520

Parity 0 1 2 3 4 5+ Plannins Status Completely planned Excess fertility

In 1960 the difference between the percentages of whites and nonwhites with contraceptive operations appeared considerable—the per­ centages were 6 and 3, respectively (Table V-24). In 1965 the esti­ mates are 7 and 8 percent, respectively. If real, this is an interesting change, but differences in the composition of the nonwhite samples in 1960 and 1965 suggest caution in interpreting this finding. The increase in contraceptive operations appears to be distributed fairly uniformly throughout the population by religion and by education. As in 1960, the population in the western part of the United States re­ mains the one with the highest proportion of such operations in 1965. (One-half of all vasectomies reported for the United States were found in the West.) This difference cannot be explained by regional variations in religion or education. The reasons are not obvious; one speculation is that doctors in that part of the country, for whatever reason, may sim­ ply be more willing to perform such operations.

CONTRACEPTION Table V-24. Percent of couples, wife 18-39, with contraceptive operations, by race and, for white couples, by region of residence, wife's religion and wife's education: 1960 and 1965.

Percent with Contraceptive Operations 1960 1965

Number of Couples 1965 1960

5 3 6

8 8 7

2684 270 2414

3756 837 2919

2 4 7 10

4 5 8 15

581 715 744 374

670 896 854 499

ReligionProtestant Catholic

8 2

9 4

1596 668

1908 846

Education College High School 4 High School 1-3 Grade School

3 5 8 8

6 7 10 9

427 1153 579 255

858 1424 643 267

Total Nonwhite White Region Northeast North Central South West

Also, as in 1960, contraceptive operations in 1965 are much more frequent among Protestants than among Catholics, and are negatively related to education. As an explanation for the higher rates of steriliza­ tion in the lower educational categories it has been suggested that doctors may recommend such operations when they believe that pregnancy would endanger the wife's health and that the couple are not effective contraceptors.18 In addition, the lower educational categories have higher proportions with excess fertility, and these are much more likely than the rest of the population to resort to sterilization. CONTRACEPTIVE EFFECTIVENESS AND DESIRED FAMILY SIZE

The use of contraception, the regularity with which it is employed, the method selected, and the resort to contraceptive surgery are all related in varying degrees to a couple's motivation to prevent a pregnancy. The degree to which contraceptive efficacy is related to the motivation to re­ strict family size was first analyzed in the Princeton Fertility Study.19 18

Whelpton et al., Fertility and Family Planning in the United States, p. 149. All three phases of the study have now been published: see C. F. Westoff, R. G. Potter, P. C. Sagi, and E. G. Mishler, Family Growth in Metropolitan Amer19

CHAPTER V

That study—of the fertility of a sample of mothers of two children ini­ tially interviewed in 1957, interviewed again in 1960, and reinterviewed for the third and final time between 1963 and 1967 as they neared the end of their reproductive lives—has provided us with the important find­ ing that the effectiveness with which contraception is practiced improves sharply as a couple achieves their desired family size.20 The inference is that motivation to control fertility becomes much greater when the desire is to stop further childbearing completely rather than simply to plan the timing of the next birth. The mechanism through which this improve­ ment occurs seems to be increased regularity of contraceptive use. The authors of the 1960 GAF Study replicated this analysis21 and found strong support for the Princeton Study findings. In 1965 we had still another opportunity to assess the relationship using data somewhat better suited to this purpose than those collected in 1960 but neverthe­ less not as suitable as those from the Princeton Study. The limitation on replicating the Princeton Study finding is not the quality or detail of in­ formation collected but rather the retrospective nature of cross-sectional studies in contrast to the "before and after" measurement inherent in the longitudinal design. Specifically, the 1960 and 1965 Studies are forced to rely on the post factum answers of women of all parities to the ques­ tion of how many children they want, whereas the longitudinal study permitted the assessment of this preference (for women with two chil­ dren) prior to additional exposure to childbearing, and then followed their subsequent histories. This before and after measurement avoids the tendency for the number of children wanted to be adjusted to correspond with actual fertility. The data assembled in Table V-25 are in the form of contraceptive failure rates, defined as the number of contraceptive failures (per 100 years of exposure) occurring in the first year of the specified interpregnancy interval. Only intervals in which contraception was practiced are considered. A failure is defined as any pregnancy reported by the woman to have occurred other than when contraception was deliberately inter­ rupted in order to conceive; as such, it includes conceptions occurring either while a contraceptive method was being used or when a method tea (Princeton: Princeton University Press, 1961); C. F. Westoff, R. G. Potter, and P. C. Sagi, The Third Child (Princeton: Princeton University Press, 1963); L. Bumpass and C. F. Westoff, The Later Years of Childbearing (Princeton: Prince­ ton University Press, 1970). 2° Westoff et al., The Third Child, pp. 38-44. The data presented there relate to the three-year experience following the birth of the second child. The same is found for data from the third interview (conducted 7 to 11 years after the first). 21 Whelpton et al., Fertility and Family Planning in the United States, pp. 288-98.

CONTRACEPTION Table V-25.

Interval

1st - 5th 1st 2nd 3rd 4th and 5th

a

Number of contraceptive failures (per 100 years of exposure) occurring in the first year of the specified pregnancy intervala for women under 45, by number of children desired.

Total

32 50 35 26 25

Number of Children Desired 0-2 5+ 3 4 .

26 39 25 20 22

25 44 30 19 13

38 72 45 33 26

52 81 61 51 43

One month for postpartum amenorrhea allowed after all pregnancies. Premarital pregnancies are excluded.

had been omitted for any reason other than to become pregnant. In both cases the woman replied that she "did not want to become pregnant at that time." "Exposure time" is defined as the first twelve months follow­ ing the termination of a pregnancy or, if never pregnant, following mar­ riage, of an interval during which the woman reported using contracep­ tion at least part of the time.22 One month was allowed for postpartum amenorrhea. Accordingly, ex­ posure time was calculated beginning one month after the termination of a pregnancy and ending either at 12 months, with an unplanned con­ ception, or with the number of months of contraceptive practice preced­ ing deliberate interruption to become pregnant,23 whichever of these events occurred first. Failure rates were computed only for the first 12 months of exposure in any interval in order to minimize the bias that results from the fact that failure rates can be affected by the length of the intended interval. If the couples in the sample who plan fertility success­ fully happen to prefer short intervals they will contribute fewer years of successful contraceptive exposure and bias the failure rates for the entire group upward; conversely, if the successful contraceptors prefer long in­ tervals they will have the opposite effect. Restricting the period of obser­ vation to the initial twelve months of exposure to the risk of pregnancy reduces this potential bias. (See Chapter XI.) The results appear in Table V-25. Failure rates for each pregnancy interval are shown to increase with the number of children desired. Fail22 Our data do not permit determination of the specific months of contraceptive use but only whether contraception was used in the interval and the contraceptive circumstances at the time of conception. 23This was calculated from replies to the question: "How many months after you stopped using a method did you become pregnant?"

CHAPTER V

ure rates also decrease at each successive pregnancy interval at all values of desired parity. Finally there is some tendency, although not as pro­ nounced as in the Princeton Fertility Study, for the failure rates to diminish as desired family size is approached. As noted above, these data are far from ideal in testing this association because the number of chil­ dren desired is measured after rather than before the fertility experience represented. Thus there is an opportunity for women to adjust their preferences to their past performance. Nevertheless, despite deficiencies in the data, it is apparent that the failure rate drops sharply as a couple approaches (and exceeds) their desired family size. SUMMARY

On the basis of data collected from three national sample surveys in 1955, 1960, and 1965, trends in attitudes toward fertility control, in the use of contraception, and in voluntary sterilization have been analyzed. The 1965 data have been further analyzed in terms of factors affecting the use of contraception, methods used currently, women's knowledge of the ovulatory cycle, and the relationship between the achievement of desired family size and the improvement of contraceptive practice. The major findings in connection with changes in attitudes since 1960 are: (1) increasing numbers of American women are in favor of the principle of fertility control; (2) the greatest change has occurred among Catholic women, many of whom have moved away from exclusive en­ dorsement of the rhythm method; (3) this change in the Catholic atti­ tude has been especially marked among the better educated Catholic women; and (4) the gap between white and nonwhite attitudes had nar­ rowed considerably by 1965 because of the rapid change in nonwhite attitudes associated with increasing education. Analysis of trends in the proportions of women who report ever having used contraception and those who expect to use it leads to the following conclusions: (1) the upward trend evident between 1955 and 1960 has continued through 1965, though necessarily at a reduced rate; (2) couples appear to be adopting contraception earlier in marriage; (3) Protestant-Catholic differences in use are continuing to diminish; (4) use of contraception has increased most sharply among the more educated Catholic women; (5) education is generally becoming less important in differentiating use; and, (6) due to a substantial increase in use among nonwhite women, especially young women in the South, the white and nonwhite differences in proportions using contraception will probably disappear in the near future.

CONTRACEPTION

A major change in the methods of contraception used by American couples occurred between 1960 and 1965. By 1965, the pill had become the most popular contraceptive method currently in use. Even among Catholics, where rhythm still predominated in 1965, the pill was the sec­ ond most popular method. Among younger women, however, the pill is the overwhelming favorite even among Catholics. As a result, the use of the diaphragm, condom, and rhythm method declined appreciably. To refine the measurement of contraceptive use, we developed a new index, the percent of couples currently at risk not using contraception. The concept "at risk" means that whether or not a couple is using a method of contraception they are vulnerable to pregnancy (i.e., they are neither pregnant, nor trying to become pregnant, nor postpartum, nor subfecund). Negro couples show a higher percentage on this index than white couples and Catholics a higher percentage than non-Catholics. We also examined group differences in types of current exposure to the risk of conception and methods of contraception currently used. Negro couples are more likely to interrupt contraceptive practice without any explicit or obvious fertility objectives. In the 1965 National Fertility Study questions were also asked to determine the accuracy of women's knowledge about the biology of the ovulatory cycle. Twice as high a proportion of white women as of Negro women are classified as "correctly informed," while Catholic women, who rely extensively on the rhythm method, are slightly better informed than non-Catholic women. The amount of general education a woman has is clearly a factor in her knowledge of such matters. There appears to be a trend toward greater reliance on surgical steri­ lization (both male and female) for contraceptive reasons, although the overall magnitude is still small. The relatively high incidence in the West, and within the lower educational classes, as observed in 1960, is still apparent in 1965. Analysis of the significance of desired family size for the effectiveness of contraceptive practice, first explored in the Princeton Fertility Study and replicated in part in the 1960 GAF Study, was again tested on the 1965 data, with approximately the same outcome. Contraceptive failure rates drop markedly as a couple nears (or exceeds) desired family size. As has been noted in the earlier studies, this finding highlights the im­ portance of motivation for the effectiveness of contraceptive practice, and accordingly carries numerous implications for understanding differ­ ences in failure rate by parity, by method, and by religion.

CHAPTER VI

Use of the Pill THERE are many respects in which oral contraception represents a critical area for research. It exploits a new principle in fertility regula­ tion, the separation of the processes of copulation and procreation; it was the focus of recent debate within the Roman Catholic Church about the morality of various means for achieving responsible parenthood; and it is suspected of playing a major role in the recent decline of American fertility. Moreover, contraceptive technology is experiencing yet another transformation with the intrauterine contraceptive devices. Although manufacturers had released estimates of the volume of sales, it was data from the 1965 Study that made possible the first systematic assessment of the extent of the acceptance and use of oral contraception by the total population of married couples1 and that permitted various demographic and social analyses. The relationship between the increasing adoption of the pill and the declining fertility of American women is also explored2 in this chapter. DIFFUSION In order to capture the early history of the diffusion of the oral con­ traceptive, all women who reported any use of the pill were asked in which months since January 1960 they had used it. These data permitted the calculation, month by month, of the proportion of women using the pill—a unique time series of considerable interest not only in relation to fertility movements per se, but also to the broader study of the diffusion of innovations. The proportion of our sample using oral contraceptives for the months 1 Our first publication based on data from the 1965 National Fertility Study was devoted to an estimate of the extent of use of the pill and the characteristics of users. Much of the material in the beginning part of this chapter is drawn from that article: Ν. B. Ryder and C. F. Westoif, "Use of Oral Contraception in the United States, 1965," Science 153 (Sept. 9, 1966): 1199-205. See also C. F. Westoffi and Ν. B. Ryder, "Experience with Oral Contraception in the United States, 19601965," Clinical Obstetrics and Gynecology, vol. 11, no. 3, New York: Harper Medical Division of Harper and Row (September, 1968): 734-52. That article summarized some of the material in the Science article as well as highlights from those cited below. 2 Ν. B. Ryder and C. F. Westoff, "Oral Contraception and American Birth Rates," in W. T. Liu, ed. Family and Fertility (Notre Dame, Indiana: University of Notre Dame Press, 1967), pp. 171-83. Also published as "The United States: The Pill and the Birth Rate, 1960-1965," Studies in Family Planning, no. 20, The Population Council (June 1967): 1-3.

CHAPTER Vl

of January 1960 through September 1965 are depicted in Figure 1. This gives the percentage of women, born since July 1, 1920, and first married by the month concerned, who reported that they used the pill in that month. The series shows uninterrupted increase by an increasing amount per month. Words are scarcely necessary to convey the demographic and Figure 1.

Percent of married women (husband present), born since mid-1920 currently using oral contraceptives: United States, January, 1960 - September, 1965·

PERCENT

16.0 15.0 14.0 13.0 12.0

0.0 9.0

8.0 7.0 6.0 5.0 4.0 3.0 2.0

0.0

I960

1961

1963

1962

1964

1965

social interest of this time series. Separate time series for women first married since I960, when the pill was licensed for contraceptive pre­ scription, and for women first married before that time, presented in Table VI-1, show that use is much more common among the more re­ cently married wives, but that it is increasing rapidly in both marriageduration categories. Table VI-l.

Duration

Percent of -women using oral contraception, by marital- duration, for the years 1961-65.

1961

1962

1963

1964

1965

0-4 Years"

3.9

5.5

10.Φ

15.1

25.6

5+ Years

0.7

1.7

3.71

7.3

11.1

USE OF THE PILL

EXTENT OF USE

The patterns of pill use are summarized in six categories for purposes of analysis. All women who had "ever used" the oral contraceptive are divided into those "now using," those who indicated that they "will not use again" in the future, and those who "may use again." The women reporting they had "never used" consist of those who say they "will not use" in the future, those who "may use," and those who have "never heard of it." Now using. This category comprises the 15 percent of all married women under age 45 who reported that they were currently using the oral contraceptive. May use again. Six percent of married women fell into this category. They reported, on a month-by-month time chart covering the period 1960 to 1965, that they had used contraceptive pills but that they were not currently using them. When asked, "Do you think you might use them in the future?" they answered either "Yes" (4 percent) or "Don't know" (2 percent). We decided to combine these two responses after determining, by analyz­ ing them separately, that they were not related differentially to the other variables in our analysis. It is our opinion that most of the women in this category are either pregnant or trying to become so. Will not use again. Five percent of all married women had discon­ tinued use of the oral contraceptive and answered "No" to the ques­ tion about future use. Collectively, these first three categories constitute the 26 percent of all married women under age 45 who have ever used the oral contracep­ tive. The remaining 74 percent who report no use are divided into the following three categories: May use. Nineteen percent of the women in our sample replied nega­ tively to the question about past use, but answered "Yes" (11 per­ cent) or "Don't know" (8 percent) to the question about future use. As with the second category, we decided to combine these two re­ sponses because separate analysis of the two failed to reveal differ­ ences in relation to other variables. Will not use. These women, 52 percent of the total, have never used oral contraception and answered "No" when asked "Do you think you might use it in the future?"

CHAPTER Vl

Never heard of it. The final category consists of 3 percent of all mar­ ried women who responded "No" when asked the initial question in the series, "Have you ever heard about the oral contraceptive—the birth control pill?" Table VI-2 presents estimates of the number of married women in the United States, on the basis of the proportion determined for the sample, who fall into each of the aforementioned six categories. It is estimated that 3.8 million married women, under 45 and living with their husbands, were using the oral contraceptive in 1965. This estimate is subject to sampling error; the probability is 0.95 that the true value lies between 3.5 and 4.1 million. We also note from Table VI-2 that 6.4 million married women have used the oral contraceptive at some time and an additional 4.7 million Iable VI-2. Estimated number (in thousands) and percent of married women in the United States in 1965 who had used, were using, or expected to use oral contraception, by age.

Under 45

Under 20

20-24

25-29

30-34

35-44

Number (in thousands) Total

24,645

905

4,065

4,646

4,736 10,293

Ever Used, Total Now using May use again Will not use again

6,388 3,815 1,341 1,232

389 258 89 42

1,906 1,241 419 246

1,726 1,042 373 311

1,160 628 206 326

1,207 646 254 307

Never Used, Total May use Will not use Never heard of it

18,257 4,676 12,794 • 787

516 261 .242 13

2,159 852 1,188 119

2,920 1,050 1,795 75

3,576 931 2,502 143

9,086 1,582 7,067 437

100

100

100

100

100

Percent Total

100

Ever Used, Total Now using May use again Will not use again

25.9 15.5 5.4 5.0

43.0 28.5 9.8 4.6

46.9 30.5 10.3 6.1

37.2 22.4 8.0 6.7

24.5 13.3 4.3 6.9

11. 6. 2. 3.1

Never Used, Total May use Will not use NeJrer heard of it

74.1 19.0 51.9 3.2

57.0 28.8 26.7 1.4

53.1 21.0 29.2 2.9

62.8 22.6 38.6 1.6

75.5 19.7 52.8 3.0

88. 15.· 68. 4.:

Source: U. S. Bureau of the Census, "Population Characteristics," Current Population Studies, P-20, no. 144 (November 10, 1965), Table 1.

USE OF THE PILL

indicate that they may use it in the future. Large though these figures on use are, it should be emphasized that they are underestimates of the total experience with oral contraception since 1960. We have excluded from this analysis women who are beyond age 45. Furthermore, our sample does not provide data on use by women who are in marital statuses other than currently married and living with husband. Accordingly, we have no data on past or present use by women who have never been married or by women whose marriages have been interrupted for some reason. Correct interpretation of these data also requires recognition of the fact that not all use is for contraceptive purposes. Oral progestins are also used for the relief of menstrual discomforts, for regularization of the ovulatory cycle—frequently with the intent of promoting a subsequent conception—and for many other purely medical purposes. We asked women who had used the pill whether the use was "at least partly to delay or prevent a pregnancy." One-sixth of these women answered in the negative; they constitute some 4.3 percent of all married women in the age group. Eighty-three percent of married women who have used oral progestins have used them at least in part for purposes of contra­ ception. Use exclusively for purposes other than contraception is con­ centrated among women who have stopped using the pill (30 percent) in contrast to current users (7 percent). Use for purposes other than contraception is reported by 1 percent of all Negroes but by 5 percent of all whites. We believe that this difference reflects comparable disparities in medical care by race. It is unlikely that such disparities can explain the difference in percentages for white nonCatholics (4 percent of whom report use only for reasons other than contraception) and white Catholics (6 percent). Some of this difference may reflect temporary use of the oral contraceptive by Catholic women to regularize the ovulatory cycle and improve the efficacy with which they can employ the rhythm method of fertility regulation. We suspect, however, that part of the explanation may stem from a greater tendency toward dissimulation among Catholics on a question that is of great moral concern to many of them. USE BY AGE AND PARITY

Table VI-3 presents use, measured in three ways, for the total sample crossclassified by age and parity. In this and later tables, "now using" refers to the first category in the code described above, "ever used" re­ fers to the first three categories, and "have used or may use" refers to the first four categories. It is apparent, on all three bases of measurement,

CHAPTER Vl

Table VI-3.

Use of oral contraception, by age and parity.

Parity 2 3

Age

0

1

Total

14

16

18

Under 20 20-24 25-29 30-34 35-39 40-44

19 28 13 3 0 2

30 26 15 10 5 1

49 37 25 14 9 •5

4

5+

Total

15

17

11

15

30 24. 15 7 8

32 30 14 12 6

18 23 14 9 4

28 30 22 13 8 5

Now Usins (%)

Ever Used (%) Total

26

29

27

26

26

20

26

Under 20 20-24 25-29 30-34 35-39 40-44

29 45 31 14 4 3

47 44 32 24 12 5

59 52 35 26 14 8

44 39 24 15 15

43 48 24 22 9

36 34 28 16 7

36 42 37 24 15 9

Total

48

49

47

44

44

37

45

Under 20 20-24 25-29 30-34 35-39 40-44

65 66 60 37 17 U

76 63 56 46 27 15

71 73 58 51 36 19

69 61 44 33 26

70 70 40 42 20

67 54 42 39 21

70 68 60 44 34 20

Total

549

767

1177

975

589

751

4808

Under 20 20-24 25-29 30-34 35-39 40-44

89 171 74 70 75 70

79 242 125 99 93 129

32 227 234 216 214 254

11 124 208 219 212 201

1 64 108 156 129 131

0 26 101 204 200 220

212 854 850 964 923 1005

Have Used or May Use (X).

Number of Women

that the use of oral contraception is much greater among younger than among older women. Of married women under 30, more than two-fifths have already used the pill; of married women over 40, less than onetenth have used it. The dominant position already achieved by the oral contraceptive among young couples is suggested by the observation that, whereas in 1960 only 75 percent of couples where the wife was under 25

USE OF THE PILL

had ever used any method of contraception, by 1965 45 percent of same age group had used the pill. A number of circumstances explain the observed variations in use by age. These include such factors as: the decline of use with advancing age because of the gradual onset of sterility; the competition with the oral contraceptive of other methods which older couples may have found to be satisfactory regulatory procedures; the tendency for younger cohorts to be more receptive to any innovation; and, in the case of young married couples, the particular appeal of a procedure for fertility control which may enhance rather than inhibit sexual satisfaction. The relation of use to age is not uniformly monotonic—the propor­ tions in the "now using" and "ever used" categories are somewhat smaller for married women under 20 than for those between 20 and 24. Three explanations may be suggested for this finding. In the first place, the circumstances predisposing to early marriage may also inhibit prompt use of contraception. Secondly, at the time of interview these marriages are of shorter average duration than those of somewhat older women; accordingly, there has been less opportunity for use of contraception. Finally, use is expected to rise with increasing age as the numbers who use the pill for timing births are augmented by those who use it for terminating fertility. This interpretation is supported by data on the per­ centages of married women who contemplate possible future use of the pill, which show a maximum of 70 percent for the youngest cohort. The superficial expectation that the use of the pill should be closely related to parity is not supported by the data. Variations in use with dif­ ferences in number of live births are small. It is likely that this result reflects several crosscurrents of influence. Some women use oral contra­ ception for timing the births of their first few children, while others use it to terminate childbearing when they have the desired number of chil­ dren, a number which differs from couple to couple. Still other women may start using the pill after they have already had more than the num­ ber of children desired, perhaps because other methods have proved un­ successful. Also, the lowest parities in a cross-sectional sample like this tend to have a disproportionately large number of sterile women. Finally, women in the highest parity groups may use oral contraception less be­ cause, for reasons of race, religion, or education, they have less effective access to information about contraception. By age and parity combined, Table VI-3 shows that the greatest current use is among women who are still young and who have two children. Current use, not surprisingly, is

CHAPTER VI

least common among women of advanced age in the lowest parity groups. USE BY EDUCATION

Table VI-4 presents data reflecting a variable of major relevance for the adoption of a new method of contraception: the number of years of schooling a woman has completed. The percentages of married women Table VI-4. Use of oral contraception, by wife's age and education.

Education Grade H.S. 1-3 School

College 4+

College 1-3

Total

22

19

16

15

7

15

Under 20 20-24 25-29 30-34 35-39 40-44

47 33 20 8 10

41 26 14 8 7

31 31 20 14 7 6

27 24 21 11 11 3

27 15 21 7 3 2

28 30 22 13 8 5

Ever Used W 26 26

12

26

42 40 36 20 22 6

43 28 27 12 5 2

42 46 37 24 15

28

45

69 67 55 35 34 18

73 52 53 29 20 11

70 68 60 44 34 20

Number of Women 2106 1214

630

4807

96 253 207 202 207 249

21 57 70 131 152 199

212 854 850 963 923 1005

Age

H.S. 4

Now Using

Total

37

32

Under 20 20-24 25-29 30-34 35-39 40-44

71 52 36 22 15

61 37 34 16 12

Total

57

Under 20 20-24 25-29 30-34 35-39 40-44

78 74 63 44 30

75 78 60 58 40 21

318

539

0 39 70 81 64 64

12 105 110 118 93 101

Total Under 20 20-24 25-29 30-34 35-39 40-44

45 46 36 24 13 11

«

Have Used or May Use (%) 52 44 47 70 67 60 45 36 23

83 400 393 431 407 392

Total

9

USE OF THE PILL

now using oral contraception vary positively and strongly with the level of education—from 22 percent for college graduates down to 7 percent for individuals with only a grade school education. The same strong re­ lationship holds for past and prospective use and is independent of age. If the variations by education were less for the "ever used" category than for the "have used or may use" category, this might imply that education influences the time of adopting the new method rather than the likeli­ hood of eventual use. Such an interpretation is not warranted by the data at hand, although it is likely that many couples who have not yet used the oral contraceptive will subsequently change their opinion on future use. The education-related differences observable in Table VI-4 probably reflect the greater awareness of and receptivity to innovation that are generally cultivated by education. The proportion of women who say they have never heard of the oral contraceptive is only 2 percent for those who at least reached high school but is more than 14 percent for those whose education terminated in grade school. The importance of the data on use by age and education is that they provide the best indi­ cation of future use, as newer cohorts replace older ones and the pro­ portions of women with more education increase. The majority of younger married women with at least some college education has already used the oral contraceptive; statements on prospective use suggest that eventually three-quarters of these women will do so. USE BY WHITES AND NEGROES

The data in Table VI-5 for whites and Negroes show less extensive use by Negroes and reveal different relationships for the two races between use and age and parity, but not between use and education. There is con­ siderable evidence here for the view that Negroes are using oral contra­ ception at a later stage in their marriages than whites, presumably to terminate rather than to time fertility. Thus, by age, the percentage of women now using oral contraception is markedly different for whites and Negroes under 25, but the divergence is trivial for ages 25-29 and 30-34. Similarly, the percentage of women currently using the pill is much higher for whites than for Negroes in parities 0 and 1, but the posi­ tions of the races tend to reverse in higher parities. As for the "have used or may use" category, the only large discrepancy between whites and Negroes is in the youngest age and lowest parity groups. Fewer differ­ ences, by race, are found in the education section of Table VI-5 than are found in the age and parity divisions. In other words, much, although not

CHAPTER Vl Table VI-5.

Use of oral contraception by whites (W) and Negroes (N), by age, parity, and education.

Percent Now Using W N

Ever Used N W

Have Used or May Use W N

30 32 22 14 8 5

17 17 23 12 5 2

45 49 38 25 15 9

22 30 25 18 9 9

74 68 60 44 35 20

44 55 56 46 29 12

155 640 650 762 741 821

54 203 183 185 172 172

16 17 18 16 16 10

4 10 14 14 23 11

29 31 28 26 26 20

8 19 22 23 29 15

52 49 47 44 44 36

24 43 50 46 54 40

426 604 987 797 483 473

119 153 173 163 99 262

22 19 16 16 9

21 16 17 12 3

38 33 26 27 13

29 24 25 19 6

57 53 47 43 28

50 46 49 45 27

277 457 1788 850 396

38 74 302 343 212

16

12

27

19

45

43

3769

969

Number of Women W N

Age Under 20 20-24 25-29 30-34 35-39 40-44 Parity 0 1 2 3 4 5+ Education College 4+ College 1-3 High School 4 High School 1-3 Grade School Total

all, of the racial difference in the extent to which oral contraception is used is explainable by differences in education. A comparison of the percentages in the "now using" and "ever used" cnlnmns shows the differences between the races to be smaller for the "now using" category. Two possible explanations for this may be sug­ gested. On the one hand, such a contrast would be expected if a Negro were less likely than a white to use the pill for timing early fertility and more likely to use it for terminating fertility. This would produce a larger difference in the "ever used" than in the "now using" category, because use for terminating fertility is less likely to be interrupted than use for timing births. On the other hand, it is possible that Negroes did not learn of the availability of the oral contraceptive as early as whites; such a cir­ cumstance would also produce the contrast noted. Ten percent of Negroes have never heard of the oral contraceptive, whereas the com-

USE OF THE PILL

parable figure for whites is less than 3 percent. Tabulations by region show that this discrepancy is a consequence of a large difference in the South. For Negroes in the South, the proportion who have never heard of the oral contraceptive is 14 percent; for Negroes elsewhere in the na­ tion, the proportion is essentially the same as for all whites (3 percent). USE BY RELIGION

The recent history of debate within the Catholic Church on contra­ ception is well known even to a casual reader of the daily press. Develop­ ment of the oral contraceptive created major disagreements within the Catholic Church over its moral acceptability, in part because it differs from conventional contraceptives in its modus operandi and in part be­ cause it appeared on the scene during a period of mounting concern over the problems created by rapid population growth. During this debate, a great deal of curiosity and speculation arose concerning the extent to which Catholic women in the United States do in fact use contraceptives. Apparently some parish priests decided to declare the oral contraceptive acceptable; evidently many Catholic couples decided either to proceed on the assumption that the oral contraceptive would be declared accepta­ ble by the Catholic Church or to ignore injunctions against its use. Since other major religious denominations in the United States do not take any strong position on the use of oral contraception, we have di­ vided the sample of the white population into Catholic and non-Catholic; within the non-Catholic category, the principal components are Prot­ estants and Jews, and they do not differ in their use of oral contraception. The data in Table VI-6 reveal the differences in use of the oral contra­ ceptive by Catholics and non-Catholics. The most noteworthy feature of the table is the extent of use it indicates among the Catholic women in the survey. Twenty-one percent of the Catholic women under the age of 45 have used the oral contraceptive3 (and 29 percent of the non-Catho­ lic women). As noted above, a larger proportion of Catholics than nonCatholics report that the use is for medical reasons. Excluding these, we find that the percentages of women who have used oral progestins at least in part to regulate fertility are as follows: Catholics, 15 percent; nonCatholics, 25 percent. The largest differences in use are found in the youngest age groups and lowest parity groups. Most striking is the religious discrepancy in the under-20 age group, where 56 percent of non-Catholics but only 15 per8A considerable increase in the use of oral contraception by Catholics occurred between 1965 and 1969. (See Chapter VIII.)

CHAPTER Vl Table VI-6.

Use of oral contraception, by age and education, by white non-Catholics (WNC) and white Catholics (WC).

Percent Now Using WNC WC

Ever Used WNC WC

Have Used or May Use WNC WC

Number of Women WNC WC

Age Under 20 20-24 25-29 30-34 35-39 40-44

37 35 24 16 9 6

10 25 18 9 5 3

56 54 39 29 15 11

15 36 35 18 16 6

75 74 60 48 35 23

56 54 61 36 35 16

114 452 474 520 535 580

41 188 176 242 206 241

19 18 20 17 19 12

8 13 13 12 10 8

35 34 29 28 .29 20

12 22 25 20 21 21

58 51 50 45 46 34

33 44 49 40 40 39

315 443 745 578 309 285

111 161 241 219 174 188

24 21 18 17 10

12 15 11 12 6

38 34 29 30 14

33 28 20 20 13

58 54 49 46 28

51 49 41 37 27

226 341 1235 604 268

51 116 553 246 128

18

11

29

21

48

39

2675

1094

Parity 0 1 2 3 4 5+ Education College 4+ College 1-3 High School 4 High School 1-3 Grade School Total.

cent of Catholics report ever having used the birth control pill. Although some of this difference may be attributable to religion-associated differ­ ences in age at marriage and in duration of marriage, the principal explanation is probably that Catholics begin use of contraception at a later stage in their marriage than non-Catholics. This interpretation is supported by the tendency for percentages of Catholics and non-Catho­ lics in the "now using" category to converge in the higher age and parity groups, and particularly by the similarity of percentages in the "have used or may use" category beyond the lowest age and parity groups. Data by age and parity (not included here) show that the largest dif­ ference in use of the pill occurs among childless women under age 25. In this category, 34 percent of the non-Catholics but only 7 percent of the Catholics are currently using oral contraception. The differentials are considerably attenuated for higher age-parity combinations. The contrast

USE OF THE PILL

is similar to that observed for whites and Negroes. Non-Catholics are likely to use the pill for timing births as well as for terminating fertility; Catholics are more likely to use it solely for terminating fertility. Prediction of future use is difficult. In the youngest age groups, Catho­ lics appear to be much less likely to use oral contraception eventually than non-Catholics. Nonetheless, in the 25-29 age group the percentage of Catholics in the "have used or may use" category is about the same as that of non-Catholics. From these data it seems reasonable to suggest that Catholic attitudes toward oral contraception are modified as families grow in size. The data on use by education do not indicate the need for any important modification of this interpretation—approximately the same patterns of difference in current, past, and prospective use are found. In both religious categories, college attendance is associated with higher percentages in each "use" category. Although the extent of use may be lower among Catholics than nonCatholics, the proportion of Catholics who report use is substantial in­ deed, in view of the theological controversy of the times. THE PILL AND THE BIRTH RATE

Given the growing adoption of a highly effective contraceptive, the critical demographic question is the extent of its connection with the downward trend of the American birth rate that began in 1957. In order to evaluate the pill's significance for the interval 1960-65, marital fertility rates for 1960 and 1965 were estimated by dividing each age-specific fertility rate by the corresponding proportion of women mar­ ried, husband present. The 1965 age pattern of pill use was assumed for the very small number of users in 1960. Table VI-7 shows the percent Table VX»7.

Changes in marital fertility and pill usa between 1960 and 1965.

Percent Decline in Marital Fertility

Increase in Percent Using Pill

14-19 20-24 25-29 30-34 35-44

8 20 17 15

17 19 14

18

4

14-44

19

10

Age Group

8

CHAPTER VI

decline in marital fertility between 1960 and 1965 and the increase in the proportion of married women using the pill (with a 9-month lead). Examination of the two proportions for each age group makes it appar­ ent that the age pattern of decline is quite unlike the age pattern of pill adoption. Although the employment of oral contraception is most evi­ dent in the youngest age groups, the decline in fertility has not been con­ fined to these ages. A similar conclusion emerges when yearly increase in pill use (again with a 9-month lead) is compared to the decline in to­ tal fertility in Table VI-8. The decline in total fertility is well in advance of the increase in pill use. (It should be remembered that total fertility also reflects changes in the proportion of women married, husband present.) Table VI-8.

Interval

Year-by-year changes in total fertility and pill use between 1960 and 1965.

Percent Decline in Total Fertility

Increase in Percent Using Pill

1960-61

1.0

0.3

1961-62

4.0

0.5

1962-63

4.2

1.7

1963-64

4.0

3.6

1964-65

8.4

4.6

It is hoped that, eventually, through more sophisticated analysis, a clear link may be established between the contraceptive behavior and the reproduction of American couples. The significance of the pill lies not only in its effect of enhancing the ability to terminate childbearing, but also in its use to delay the births of the first few children. Whether or not a change in eventual family size occurs, an increasing tendency to delay would lead to a decline in annual birth statistics. It is also Ukely that couples who delay the birth of their first and second children will have smaller families than those who do not. In advance of that necessary analysis, it does seem pertinent to note that the oral contraceptive is only one of a battery of procedures which have been employed successfully over a long time, that in late 1965 the majority of women were still em­ ploying other contraceptive procedures, and that the use of the pill is very recent. During the Depression, the American people demonstrated

USE OF THE PILL

their ability to use very effectively means of fertility regulation which had been known for a long time. And during the postwar baby boom the population demonstrated that, in the area of reproduction control, the ends as well as the means are highly variable, and that for this reason de­ tailed investigation of the voluntary component in childbearing is essen­ tial for rigorous analysis. It is likely that the direction, if not the degree, of change in marital fertility in the 1960's would have been the same even if the oral contra­ ceptive had not appeared on the scene. However, the tempo of decline at mid-decade can probably be attributed in part to the availability of this highly efficient and highly acceptable method of fertility regulation. Its contribution to fertility decline among recently married couples is already substantial. The young American wife has clearly shown an extraordinary enthusiasm for oral contraception. In the 1965 National Fertility Study, the category of white, non-Catholic women, 20-24 years old, with college diplomas yields the highest proportion of women who have ever used the oral contraceptive: 81 percent, and an additional 4 percent who say they may use it in the future. This is indeed an extraor­ dinary level of acceptance for a procedure which was not even available until 1960. Increasing educational levels and improvements in the product itself would undoubtedly increase its use and acceptance. The most reasonable prediction in the developing field of fertility regulation is that there will be change; the emergence of new products which alter the entire pattern of contraceptive practice is the most likely possibility. Yet, it is evident from the data presented here that the oral contracep­ tive, in the few years in which it has been available, has succeeded in transforming the character of fertility regulation in the United States. Five years after its inception, the pill was already the leading method of contraception among American women.

CHAPTER VII

Some Effects of Pill Use EVER since the pill was licensed for sale in the United States there has been an active and occasionally vociferous concern expressed about its long-term effects on the health of women using it. Although the data ob­ tained in the 1965 National Fertility Study were in no sense designed to evaluate this matter, the information collected does permit estimating the length of time women have used the pill, the probability of discontinuing the method, and the reasons women report for discontinuation. We are also able to determine which methods of contraception are used after the pill is discontinued. These are the subjects discussed in the first part of this chapter.1 The second half of the chapter is devoted to an examination of two different kinds of possible effects of the pill, both of which have received some attention in the popular and scientific press. The first is the interest­ ing question of whether married women who use the pill experience a higher coital frequency than women who use other contraceptive methods. The second question is whether women who have used the pill conceive more rapidly after discontinuation than do women who inter­ rupt other methods of contraception in order to conceive. In attempting to answer both of these questions, which incidentally are burdened with many methodological difficulties, we have assembled evidence from the longitudinal Princeton Fertility Study as well as from the 1965 National Fertility Study.2 DISCONTINUATION The extent to which women discontinue oral contraception and their reasons for discontinuing are of interest for two major reasons. First, since the pill is the most effective contraceptive yet developed, there is a high demand for information about its acceptability from persons con­ cerned with fertility-planning programs in the United States and else­ where. Second, the pill is more than just another contraceptive—it is a distinctive type of method. Ever since it was licensed for sale in the 1 This material was first published in C. F. Westoff and Ν. B. Ryder, "Duration of Use of Oral Contraception in the United States, 1960-65," Public Health Reports, vol. 83, no. 4 (April, 1968): 277-87. 2 The results of this analysis were published originally by C. F. Westoff, L. Bumpass, and Ν. B. Ryder, "Oral Contraception, Coital Frequency, and the Time Required to Conceive," Social Biology, vol. 16, no. 1 (March, 1969): 1-10.

CHAPTER Vll

United States in 1960 there has been much concern about possiblejaadesirable consequences for the user's health. Although confidence in its safety has increased with time and the accumulation of satisfactory ex­ perience, the Food and Drug Administration3 as well as the drug com­ panies which market the various brands of birth control pills have con­ tinued to be watchful, especially for long-term effects. The importance of this question is enhanced by the fact that the pill is now the leading method employed by American women (see Chapter VI). Since our interview included questions concerning the respondents' experience with the pill, we were able to estimate the proportion who had used it and their reasons for use, the proportion who had discontinued use and their reasons for discontinuing, and attitudes in general toward the pill. From a calendar of month-by-month use for the period 1960-65, we estimated the probability of dropout over time. In Chapter VI, we presented estimates that 26 percent of our sample had used the pill at some time and that 40 percent of these had stopped using it. More than half (52 percent) of those who discontinued use in­ dicated that they may use it again. Obviously, this is a most unrefined estimate of discontinuation. It includes not only women who were using the pill solely for contraceptive purposes but also those who said they were using it partly or entirely for other reasons, such as to relieve men­ strual discomfort or to increase fecundability, and who may have stopped use because the pill was or was not effective for their particular prob­ lems. This unrefined proportion who had stopped using the pill also in­ cluded those who discontinued for reasons unrelated to satisfaction with the method as such; they may have stopped in order to conceive or because of the onset of sterility. Finally, sophisticated measurement re­ quires not only the fact of discontinuing but also the length of time used before discontinuing. We attempt to remedy the deficiencies described above by focusing on the women who said they were using the pill for contraceptive reasons only. Thus, we distinguish dropout for reasons associated with the pill from dropout for extraneous reasons and explore the time dimension of use by life-table procedures. Of all women who had ever used the pill, 69 percent reported use for contraceptive reasons only, 13 percent reported use for noncontraceptive reasons only, and 15 percent reported use for both types of reason. The remainder (a little more than 3 percent) said that they were using it for 3Food and Drug Administration, Advisory Committee on Obstetrics, "Report on Oral Contraceptives" (Washington, D.C.: U. S. Government Printing Office, August 1, 1966).

EFFECTS OF PILL USE

noncontraceptive reasons only, but later in the interview identified their method of contraception as the pill. Of women reporting having used the pill for contraceptive reasons only, 66 percent were using it at the time of interview; of the one-third who had discontinued, 53 percent said that they may use it again. Of those reporting use for both contraceptive and noncontraceptive reasons, 45 percent were using the pill at the time of interview; of those in this category who had discontinued use, 50 percent said they may use it again. These proportions suggest that the pill is used longer for contraception than for other reasons. The division of respondents into those using the pill for contraceptive reasons only and those using at least in part for noncontraceptive reasons is based on self-reports which, since the question carries ethical implica­ tions for some respondents, may lack validity. Thus, among white nonCatholics, one out of seven reported use at least in part for noncontra­ ceptive reasons, but among white Catholics, the proportion was twice as high. As we suggested before, some of this difference may be due to temporary use by Catholic women to regulate the ovulatory cycle, but some may also be due to a greater tendency of Catholics to rationalize contraceptive motivations. Furthermore, among the reasons for choosing the oral contraceptive in preference to other methods there may be a marginal perception of beneficial side effects, such as the avoidance of dysmenorrhea, which would be classified as mixed motivation. There is no way we can resolve these issues of validity with the data at hand. The remainder of this analysis is concerned exclusively with the women who reported that they used the pill only for contraception. Reasons for Discontinuation

Approximately one-third of the women who had used the pill were not using it when they were interviewed. These 285 women were asked: "Why did you stop using the pill?" In coding their responses, we sep­ arated them into three categories: discontinuing because of side effects; discontinuing because of other problems of use; discontinuing for reasons unassociated with characteristics of the pill.4 This categorization permits distinction between those who did and those who did not experience a problem associated with the pill (the first two categories contrasted with the third) and for those who did, whether it was physiological or not (the distinction between the first two categories). Our results were as follows: 4 R. Frank and C. Tietze, "Acceptance of an Oral Contraceptive Program in a Large Metropolitan Area," American Journal of Obstetrics and Gynecology 93 (September 1, 1965): 122-27.

CHAPTER Vll

80 percent stopped because of some problem associated with the pill— 65 percent because of side effects and 15 percent because of other diffi­ culties—and 20 percent stopped for reasons unconnected with the pill. Some women gave reasons which fell into more than one of these three categories, the most common (2 percent) being the combination of side effects and other problems of use. These women were arbitrarily assigned to the side effects category. In the few instances involving either of these two with extraneous reasons, the side effects category was used. The responses about side effects are of uneven quality—far short of the diagnostic data which might have been obtained by a medically trained investigator. Although our interviewers were experienced in gen­ eral survey research and were trained by us on the subjects of fertility and contraception, they had no special competence in the health field. The responses also reflect the expansiveness and articulateness of the re­ spondent. In the same objective situation, one woman may report many complaints, another may confine herself to the outstanding undesirable symptom, and a third may not mention any complaint. Some respondents provided precise descriptions of symptoms, others offered only vague indications of discomfort. Accordingly, our report on the distribution of responses to the question about the reason for discontinuing the pill must be evaluated with consideration of such qualifications. The principal category of reason for discontinuing was side effects: 65 percent of the women who stopped gave this as their reason for doing so. Within this category, two complaints predominated. The most fre­ quent was pregnancy-like reactions (weight change, fluid retention, breast tenderness, nausea); the next most common complaint concerned difficulties with the menstrual cycle (spotting, hemorrhaging, irregularity, cramps). Some women cited both types of symptom. The literature on oral contraception features particular concern over the possibility of side effects such as thrombo-embolic disease, blurred vision, and skin discoloration. Among the pill users in our sample, one woman reported a blood clot in one leg two weeks after discontinuing the pill, another reported being hospitalized because of a clot in one lung, and one woman reported discontinuing because she was suffering from thrombo-phlebitis. Two women reported skin discoloration; one woman reported blurry vision. We obviously are not in a position to be able to infer from these data any conclusion about the comparative gravity of the pill's consequences. Complaints of headaches and nervousness were fairly common. Al­ though the information is imprecise, it appears that the reports of nerv-

EFFECTS OF PILL USE

ousness reflect irritability and tension rather than anxiety. One woman said: "The first brand of pill I took didn't work—made me nervous." In reply to a probe about how she knew it was due to the pill, she said: "Yes, I'm sure the pills were the cause of my nervousness because I would always calm down when I went off them, during my period." An example of the fairly common complaint of headaches was: "I had such headaches I could not stand it and it made me so very nervous. I only took them for five days. I had these terrific headaches like my head was going to break apart. I just couldn't stand it; it works for some people but not for me. Doctor said to stop taking them and maybe I could try them again later but I wouldn't. We are using something else now." These two reports illustrate the crudity of our data on side effects. The list of re­ ported symptoms included many of the well-known placebo reactions— symptoms both real and imaginary as a consequence of taking dummy pills—not only headaches, nausea, and nervousness, but more specific objective reactions as well. There are obvious difficulties in inferring cause and effect. Some women may have experienced problems unrelated to the pill but coinci­ dent with their use of it. The mass media have given considerable pub­ licity to symptoms which may (or may not) be caused by the pill. Some women are presumably suggestible enough to believe they have these symptoms when the possibility is brought to their attention. A few women reported discontinuing on physician's orders, without mentioning a specific symptom. Although we have included these women in the cate­ gory of side effects, it may be that the physician's decision was merely precautionary rather than because of a specific indication. It is also evident that persons vary greatly in their ability to tolerate unpleasant side effects and in their readiness to report them. Some women in our study reported complaints such as occasional menstrual breakthroughs, but they were not sufficiently alarmed to discontinue the pill. We asked those currently using the pill the question: "What are some things that are not so good about the pill?" Thirty-eight percent of the current users reported undesirable side effects, but one-half of these (48 percent) volunteered the information that their side effects had sub­ sequently disappeared. Two-thirds of the side effects were pregnancy­ like symptoms, one-eighth were problems associated with the menstrual cycle, and one-sixth were nervousness and headaches. The second category of reasons for discontinuing use of the pill is problems other than side effects; 15 percent of the women who discon­ tinued use were so classified. Most of these problems relate to the em-

CHAPTER Vll

ployment of the pill, such as fear of forgetting to take it, anxiety that it might not work, or dislike of pills in general. One percent cited reasons of morality or religious beliefs as the basis for discontinuing. The remaining 20 percent of the women who discontinued did so for reasons unassociated with the pill itself: 15 percent stopped in order to have a child; 4 percent stopped because they no longer needed contra­ ception; 1 percent stopped for reasons on which no information was available. Although the focus of this analysis is on the decision to discontinue use of the pill, the following favorable data place in perspective the nega­ tive reactions reported above. First, as noted before, two-thirds of all users were still using the pill at time of the interview. Of those who had discontinued, one-fifth had done so for reasons unconnected with the characteristics of the pill. Furthermore, we asked all women who had dis­ continued whether they might resume use in the future. Of those who had discontinued use because of unfavorable side effects, 45 percent said they may use it again; of those who reported other problems of use, 55 per­ cent said they may use it again; and of those who reported reasons ex­ traneous to the pill, 70 percent said they may resume use. Thus, from our data, those for whom the pill was an unsatisfactory contraceptive were clearly a small minority of those who had tried it. We also asked women who had used the pill the question: "In your opinion, what are some of the good things about the pills?" Since they frequently reported more than one "good thing," the following propor­ tions add up to more than 100 percent. Predictably, 66 percent of the women noted its effectiveness as a contraceptive—and presumably most of the others thought this too obvious to mention. Thirty-two percent spoke of the feeling of security, peace of mind, or sense of well-being associated with sure avoidance of an unwanted pregnancy. Twenty-seven percent noted its convenience, ease, or simplicity of use. Twelve percent specifically commented that there were no unfavorable side effects, and thirty-five percent reported the favorable side effect that the pill reduced menstrual discomfort and irregularity. Finally, 11 percent reported that use of the pill had improved their sex relations, a report that appears to have some basis in fact, as our data on coital frequency and pill use sug­ gest. (They are reported later in this chapter.) Trend in Dropout Rate

For each woman who reported use of the pill, we compiled a monthby-month retrospective record of that use. To summarize the trend in the

EFFECTS OF PILL USE

dropout rate we assembled these data in life-table form, following the same general procedures used in mortality analysis. The probability of continuing oral contraception for at least 3, 6, 12, or 24 months is pre­ sented in Table VII-1; the probabilities reflect dropouts due to side effects only, due to side effects and other problems of use, and for all Table VII-I. Percent of women who used oral contraception for at least 3, 6, 12, or 24 months, by reason for discontinuation and year use began.

Reason and Months

1960-62

1963

1964 JulyJan.June Dec.

1965 Jan. June

Side Effects Only; 3 6 12 24

78 75 69 60

91 84 75

88 85 81

90 83

All Problems of Use: 3 6 12 24

76 70 63 53

91 82 70

88 83 78

87 77

All Reasons: 3 6 12 24

76 68 59 50

90 81 68

86 82 75

84 73

87

Number of Women

81

117

134

161

199

91

reasons, for five cohorts of women who started to use the pill in the dif­ ferent years indicated. (The definitions of the time periods employed re­ flect considerations of sample size.) Some women, of course, had more than one interval of experience with the pill because they resumed use after interruption. Our calculations are based on the respondents' most recent experience only, a decision prompted by considerations of analytical convenience and the fact that reasons for discontinuing were determined only for women not using the pill at the time of the interview. The most recent experience constitutes 87 percent of all experience. Examination of the earlier experience of women with more than one interval of use, who are of course heavily concentrated among the earlier cohorts, reveals higher dropout rates than for the more recent experience. Although it is not

CHAPTER Vll

obvious how to interpret this difference, it probably does reflect a more difficult experience with the pill in the earlier years. The higher dropout rate in the earlier years is also clearly evident in the calculations based on the most recent experience. The outstanding feature of the rates in Table VII-I is that the prob­ ability of continuing oral contraception was markedly lower for the earliest cohort than for later cohorts. There is little evidence to indicate any upward trend in the likelihood of continuation, although there is some suggestion of an increase through 1964 in the probability of con­ tinuing 1 year. Although the probability of continuing the pill for χ months is a con­ venient and easily interpretable summary, it is unsatisfactory for the comparison of interval-specific experience because it is cumulative and also because it is expressed as the complement of the topic of interest, namely the act of dropping out. We therefore prepared central dropout rates (Table VII-2) showing the dropout percent per month for the same experience described in Table VII-1. It is apparent from these cal­ culations that the low probability of continuation for the 1960-62 cohort was confined largely to the experience of the first 3 months. This relaTable VII-2.

Dropout rates (percent per month) by interval, reason for discontinuation, and year use began.

Reason and Months

1960-62

1964

1963 Jan.June

JulyDec.

1965 Jan.June

Side Effects Only 0-2 3-5 6-11 12-23

8 1 1 1

3 3 2 0

4 1 1

4 3

3

All Problems of Use 0-2 3-5 6-11 12-23

9 3 2 1

3 4 3 1

4 2 1

5 4

4

All Reasons 0-2 3-5 6-11 12-23

9 4 2 1

3 4 3 2

5 2 2

6 5

4

81

117

134

161

199

Number of Women

EFFECTS OF PILL USE

tively high dropout rate for the 1960-62 cohort is due to side effects. (The dropout rates for problems of use other than side effects can be in­ ferred by subtracting the rates in the first panel from those in the second, and so on.) Other than this, there is no systematic change over time. The reasons for the decline in the probability of discontinuation after 1962 can only be conjectured. The high dosage of the early pills has been reduced substantially. Of equal if not greater significance is the fact that physicians as well as women have become less anxious about the effects of the pill. Perhaps the same unpleasant side effects which are accepted as temporary or not serious today were cause for greater concern in the earlier years. Because of the small numbers of women in the different cohorts and the absence of any clear-cut trend in the dropout rate after the earliest experience of the 1960-62 cohort, we combined the data for all cohorts into a single synthetic cohort, as is customary in conventional life tables (Table VII-3). Thus, the experience reported for the interval of 12-23 Table VII-3.

Dropout rates (percent per month) by reasons for discontinuation and percent continuing.

Percent Continuing

Dropout Rates Interval (in Months)

0-2 3-5 6-8 9-11 12-14 15-17 18-20 21-23

Side Effects

All Problems of Use

All Reasons

4 2 1 2 1 1 1 2

5 3 1 2 1 1 1 3

5 3 2 3 1 2 1 4

Months

Side Effects

All Problems of Use

3 6 9 12 15 18 21 24

88 84 82 78 76 74 73 69

87 80 78 73 72 69 68 62

All Reasons

86 78 75 68 66 62 60 53

months is for women who began use of the pill before mid-1963; for 6-11 months, before mid-1964; and so forth. As shown in Table VII-3, 68 percent of the women were still using the pill at 1 year, and 53 per­ cent were continuing at 2 years. Considering dropouts due to side effects only, the proportions were 78 percent and 69 percent continuing at 1 and 2 years. Beyond the third month, the dropout rates for all reasons in Table VII-3 show a sawtooth-like pattern that probably results from the respondents' reporting use in rounded intervals such as 6 months or 1

CHAPTER VlI

year or 2 years. Because of this bias, the rates are assembled in Table VII-4 for intervals of 0-2, 3-5, 6-11, and 12-23 months. From the dropout rates in Table VII-3, we derived separate rates for problems of use other than side effects and for extraneous reasons (Table VII-4). Two patterns are evident: side effects are far more im­ portant than other reasons for discontinuing, at least at this stage in the history of adoption of the pill; and there is a distinct decline in the prob­ ability of dropping out because of side effects, as expected, but no such pattern for other reasons. Table VII-4.

Interval (in Months) 0-2 3-5 6-11 12-23

Dropout rates (percent per month) by separate reasons for discontinuation.

Side Effects 4 2 1 1

Other Problems of Use

0 1 0 0

Extraneous Reasons

1 0 1 1

In the following analyses of characteristics associated with discontinu­ ing use of the pill we do not consider the data for all reasons combined, because of the fundamental difference between dropping out because of side effects or other problems with the contraceptive, and dropping out in order to conceive or because there is no longer any need for contra­ ception. The latter two reasons may be explained in relation to the length of the birth interval desired, the age of the user, and other such variables; they are unrelated to the evaluation of the pill as a contraceptive. Dropout Rates by Age and Parity

As indicated in Table VII-5, women under 30 years of age (as of mid1965) showed less tendency to discontinue oral contraception than older women. Although the percentage continuing at least 2 years was higher for the younger women, this differential is produced by the large differ­ ence in the first 3 months. If the pill was used for at least 3 months, the probability of its being used for at least a year was the same for women in both age groups. For side effects, this probability is 0.88 for younger women and 0.87 for older women; for all problems of use, the corre­ sponding probabilities are 0.83 and 0.85. The greater tendency for older

EFFECTS OF PILL USE Table VII-5.

Months

Dropout rates (percent per month) and percent continuing, by age group.

Side Effects Only Under 30 30-44 Total

All Problems of Use Under 30 30-44 Total

Dropout Rate 0-2 3-5 6-11 12-23

3 2 1 1

7 3 1 1

3 3 2 1

7 3 2 2

5 3 2 X

90 83 75 67

81 75 69 56

87 80 73 62

4 2 1 1 Percent Continuing

3 6 12 24

92 88 81 73

82 76 71 63

88 84 78 69

women to discontinue use of the pill during the months immediately fol­ lowing adoption probably results from their lower tolerance of minor side effects; the younger women, especially if they have never used other methods, are more likely to be tolerant of minor physiological changes. The relationship between the parity (by mid-1965) and the probabil­ ity of discontinuing use of the pill is summarized in Table VII-6. The association with parity is similar to that with age—the higher the parity the greater the probability of discontinuation. And, as with age, the main difference is concentrated in the first interval; the probability of the Table VII-6.

Months

Dropout rates (percent per month) and percent continuing, by parity.

Side Effects Only < 3 Births 3+ Births Total

All Problems of Use < '3 B irths 3+ Births Total

Dropout Kate 0-2 3-5 6-11 12-23

3 2 2 1

5 2 1 1

90 86 79 73

87 81 76 64

4 2 1 1

4 2 2 1

5 4 2 2

5 3 2 1

85 77 70 56

.87 80 73 62

Percent Continuins 3 6 12 24

88 84 78 69

90 84 76 68

CHAPTER VII

woman with 3 or more children continuing to use the pill for at least a year if she used it for at least 3 months is no different from that of the woman with fewer children. Thus, both age and parity affect the prob­ ability of dropout mainly in the first 3 months. Since age and parity covary, and since there appears to be a difference between the dropout pattern in the first 3 months and that in later in­ tervals, we prepared Table VII-7 to show the joint influence of the two Table VII-7.

Age

Dropout rates (percent per month) for all problems of use for Intervals 0-2 and 3-24 months, by age and parity.

0-3 Months < 3 Births 3+ Births

Total

3-23 Months < 3 Births 3+ Births Total

Total

4

5

5

1

2

Z

Under 30 30-44

3 8

4 6

3 7

1 1

2 2

2 2

variables on the dropout rate. The data reveal two patterns. In the first 3-month interval, age is clearly the dominant factor and the difference by parity is mostly a reflection of the difference by age. The dropout rate beyond the third month, however, appears to be higher for women with three or more children regardless of age, although the difference may not be statistically reliable. The likelihood that a woman who has been using the pill will discon­ tinue because of problems with the contraceptive (excluding dropping out to have a child or because there is no longer need for contracep­ tion) is not simply a matter of the real or perceived physiological or psy­ chological impact that the pill has on her. Four influential variables are involved: (a) the gravity of the reaction; (b) the salience of preventing a birth; (c) the availability of alternative means of contraception; and (d) the necessity for contraception in terms of fecundity. Although it would seem, on the average, that the higher the parity the more impor­ tant the prevention of a birth, it may be that women with more children are those who have not been strongly motivated to persist with other means of contraception. Older women, moreover, and particularly those with few children, may feel that the risk of another child is smaller and less serious or they may have had relatively satisfactory experience with other contraceptives and thus be inclined to shift to them rather than per­ sist with the pill and its perceived side effects. In interpreting these data

EFFECTS OF FILL USE

it must also be considered that there is undoubtedly some degree of un­ reliability of reporting and that our samples are not large—we may be picking up some statistical static. It seems to us, however, that these data cannot be used as strong indications of the differing physiological re­ sponses by age and parity to a particular treatment without careful con­ sideration of the alternative explanations. Dropout Rates by Education and Race

The case for the importance of nonmedical reasons for discontinuing oral contraception is strengthened by consideration of the relationship between discontinuation and the amount of education a woman has re­ ceived. Table VII-8 presents interval-specific dropout rates for side Table VII-8,

Interval (in months)

Dropout rates (percent per month) and percent continuing, by education.

Side Effects Only College < H.S.4 H.S.4

All Problems of Use College H.S.4 < H.S.4

Dropout Rates 0-2 3-5 6-11 12-23

3 3 0 1

3 2 2 1

0-2 3-5 6-11 12-23

92 85 83 77

91 87 79 73

6 2 2 2

3 3 0 1

3 2 2 1

7 3 2 2

90 85 74 67

82 74 67 50

Percent Continuing 84 80 72 59

91 82 81 71

effects and for all problems of use for three educational categories: some college, 4 years of high school, and less than 4 years of high school. There is a clear division for the first interval, 0-2 months, between those who completed high school and those who did not; the latter have twice as high a dropout rate as the former. This is principally dropout due to side effects. The differences diminish beyond the third month, al­ though they tend to be in the same direction. The proportion of women "surviving" the third month who continued at least to the twelfth month is 90 percent for the higher education group, 87 percent for the middle group, and 86 percent for the low group, considering dropouts for rea­ sons of side effects only; for all problems with use, the proportions are 89 percent, 82 percent, and 82 percent.

CHATTER VIl

An interesting feature of this analysis is that college women who dis­ continue use of the pill are as likely to do so in the second 3 months as in the first 3 months; perhaps this indicates a disposition to try somewhat longer. In general, the higher dropout rate among women with less edu­ cation probably reflects the same underlying class differences in motiva­ tion and habits of planning that we think lead less-educated women to use all contraceptive methods less regularly. The differences in dropout rates between whites and nonwhites appear less than might be expected from the differences in education. The data in Table VII-9 suggest a higher dropout rate among nonwhites for prob­ lems of use other than side effects, but the differences are slight. Table VII-9.

Months

Dropout rates (percent per month) and percent continuing, by race.

Side Effects Only White Nonwhite

All Problems of Use White Nonwhite

Dropout Rates

0-2 3-5

4 2

4 1

4 3

5 3

6-11

1 1

1 1

12-23

1 1

1 3

Percent Continuing

3 6 12 24

89 84 78 70

89 85 82 66

88 81 75 66

86 79 75 52

Methods Used After the Pill

What methods of contraception do women adopt who discontinue use of the pill? Do they gravitate toward other effective methods? Are they attracted to other new methods? Do couples tend to return to methods used before the pill? For 216 women who discontinued use of the pill because of side effects or problems of use, the first method they used after the pill and the last method they used before the pill are shown in Table VII-10. A comparison of these distributions reveals the following features: ( I ) A sharp reduction, from 2 2 to 8 percent, in reliance on multiple methods, part of which could easily be an artifact of measurement. (If two or more methods had been used alternately, our questionnaire was

EFFECTS OF PILL USE Table VII-10.

Methods used before and after the pill for 216 women who had used the pill for contraceptive reasons only and who discontinued because of side effects or other problems of use.

Methoda

No Method Rhythm Douche Withdrawal Suppository Condom Diaphragm or Jelly, or Both Foam Intrauterine Device Multiple Methods Percent Total Number of Couples

a

Percent of Users Before After

24 6 6 2 1 20 16 3 0 22 100 216

30 4 5 3 1 23 13 7 5 8 100 216

Excludes women who first used the pill in combination with other methods; they comprise 7 percent of all women who discontinued use of the pill.

designed to ascertain which method had been used last only for the inter­ val since the last pregnancy. Because the time period before the pill was first used is more likely to have been preceded by a pregnancy than the period after the pill was discontinued, an artificially larger amount of multiple-method usage might have been reported for the earlier period.) (2) An increase from 24 to 30 percent in the number of women who reported using no method. Some who used no method after the pill were pregnant; if these women are excluded, the difference is erased. (3) Adoption of the intrauterine device, which was not used by any of the women before the pill, by 5 percent after they stopped the pill. The fact that fewer than 2 percent of all women who had ever used any method of contraception reported use of the intrauterine device indi­ cates that women who discontinue the pill tend to adopt other new methods. Additional evidence for this interpretation is the increase in the proportion using foam from 3 percent before to 7 percent after the pill. (4) The tendency, after discontinuing use of the pill, to return to the methods used earlier, as suggested by the similarity of proportions using the other methods listed in Table VII-10. This tendency is clearly sup­ ported by a cross-tabulation (not shown here) of the methods used be­ fore and after the pill, although the extent of the association depends on how the comparison is conceived. If all of the categories in Table VII-10

CHAPTER VlI

are used, the proportion classified the same after the pill as before is 41 percent. Excluding couples not using any method at either time, the esti­ mate is 53 percent. Finally, if couples classified in the "multiple methods" category are also excluded, the proportion using the same method at both times is 77 percent. No matter how the comparison is de­ fined, the proportion using the same method before and after the pill is substantial. THE PILL AND COITAL FREQUENCY

Two presumed effects of the oral contraceptive are its potential for enhancing the marital sexual relationship and, paradoxically, its potential for increasing fertility. Women frequently attribute improved or in­ creased marital relations to the reduction of anxiety about undesired pregnancy. Many also believe that the pill makes them physiologically more capable of conceiving after they discontinue it; indeed, the treat­ ment of sterility with the use of progesterones was an important impetus to the development of the oral contraceptive. The scientific evidence5 is very mixed and inconclusive both because of the methodological prob­ lems in measuring these effects (such as the limitation imposed by the absence of a "before and after" measurement or the lack of an adequate control group) and because of the inadequacies of the research designs and the types of samples that have been used. Although the two studies relied upon here are not focused primarily on these questions, they rep­ resent a distinct improvement over most of the small clinical studies pro­ viding the bulk of current evidence. Although it is possible that the various side effects of the pill are com­ mon enough to decrease sexual activity, the opposite point of view is usually suggested in the literature on this subject.6 There are several rea­ sons why the adoption of this modern method of contraception might result in a significant increase in marital sexual activity: the reduction of anxiety about pregnancy removes a serious impediment to coital activity; the physical separation of contraception from the sexual act may increase 5 For recent reviews of the scientific literature on these effects, see Edward E. Wallach and Celso Ramon-Garcia, "Biochemical Changes and Implications Fol­ lowing Long-Term Use of Oral Contraception," in S. J. Behrman, L. Corsa, Jr., and R. Freedman, eds., Fertility and Family Planning: A World View (Ann Arbor: University of Michigan Press, 1969), pp. 252-92. Also see E. E. Wallach and C. Ramon-Garcia, "Psychodynamic Aspects of Oral Contraception," Journal of the American Medical Association 203 (1968): 125-29. eAIthough there is, of course, a great deal of speculation about the implica­ tions of advances in contraceptive technology for premarital sexual behavior, our data are confined to married couples exclusively since they were collected as a by-product of studies of marital fertility.

EFFECTS OF PILL USE

spontaneity and enjoyment; the biochemical action of the progesterone component of the pill may increase female libido. Our data are derived from single questions in two studies addressed to the wife on the frequency of coitus with her husband during a recent period of time. From the demographic perspective, the chief interest in such matters has been the relationship between coital frequency and the chance of conception and, to a lesser extent, the study of the acceptabil­ ity of different contraceptive methods, one component of which is the frequency of coitus and other aspects of sexual behavior. Although the research objectives of the two studies used here are essentially similar, they differ in research design in ways important for our present interest. The principal difference is that the Princeton Fertility Study (PFS), is based on a longitudinal design in which women were interviewed first in 1957, six months after the birth of their second child, reinterviewed in 1960, and reinterviewed again toward the end of their reproductive period (between 1963 and 1967). The 1965 National Fertility Study (NFS), on the other hand, is a cross-sectional one-time survey of a sam­ ple of married women at all stages of childbearing. Both studies were based on probability samples of married women living with their hus­ bands, although the PFS was confined to an initial sample of 1,165 couples living in seven of the largest metropolitan areas of the United States, whereas the NFS was designed as a national sample of some 5,600 women.7 Reliability of Measurement

Our analysis begins with some findings based on the larger national sample of the NFS and concludes with data from the PFS, exploiting that study's advantage of "before" and "after" measurements of coital fre­ quency. But first a comment on the reliability of the measurement of coital frequency is appropriate. The question asked in the NFS was: "In the past four weeks, how many times have you had intercourse?" How much confidence should be placed in the reliability of such information? In order to answer such questions about many variables of interest in fer­ tility surveys, we drew a simple random subsample of 408 women from 7 The Princeton Fertility Study has been supported chiefly by the Carnegie Corporation, with supplementary grants from the Population Council. For further details on the sample and other findings of the Princeton Fertility Study, see L. Bumpass and C. F. Westoff, The Later Years of Childbearing (Princeton: Princeton University Press, 1970); C. F. Westoff, R. G. Potter, Jr., and P. C. Sagi, The Third Child (Princeton: Princeton University Press, 1963); and C. F. Westoff, R. G. Potter, Jr., P. C. Sagi, and E. G. Mishler, Family Growth in Metropolitan America (Princeton: Princeton University Press, 1961).

CHAPTER Vll

the 5,617 women initially interviewed for the NFS, and reinterviewed them approximately three months later with the same set of questions.8 (Only the responses of the 354 women under 45 are represented here.) It is clear, of course, that the time referent is different and that there could be genuine change, because of childbirth or the husband's absence, for instance, as well as because of general temporal variability. For this reason, our estimates of reliability are minimal. Perhaps it is more ap­ propriate to regard the reported measure as reflecting the stability of behavior (or the perception of behavior) as well as the reliability of measurement.9 In any event, our main concern here is whether there is any important difference between women using the pill and women using other methods of contraception in the consistency of reporting coital frequency, a difference which could affect the comparability of changes over time in the coital frequency of the two groups. Perhaps pill users, being generally better educated, report more reliably. The measures of aggregate consistency in Table VII-Il indicate a somewhat higher mean (and variance) of coital frequency reported in the second interview, a change that could of course be genuine. Contrary to our expectation, the measures of individual consistency, although low Table VII-11.

Consistency in reporting coital frequency in two inter­ views, by current use of the pill, other methods, or no method.

Statistic Ratio of Meansa Katio of Standard Deviations Correlation Coefficient Percent on Main Diagonal J3 Coefficient of Consistency Ntmiber of Women

a

Total

Pill

0.97 0.95 0.65 26 0.18 300

0.90 0.98 0.48 28 0.19 46

Method Used Other 0.97 0.91 0.78 32 0.24 122

None 1.02 0.99 0.58 20 0.12 132

The statistic from the first interview divided by the statistic from the second interview. Defined as (0 - E)/(N - E) where 0 is the frequency observed on the main diagonal and E is the frequency expected on the main diagonal. 8

See Appendix A for detailed analysis of our reliability data. factor reducing these measures of consistency is the report of frequency as a range at either one or both interviews. The way range responses were coded could result in situations in which two responses would be considered "inconsistent" by our measures even though one was completely included within the other. 9 Another

EFFECTS OF PILL VSE

in general, show higher values for women using other methods than for women using the pill. Perhaps sexual relations are remembered more accurately when connected with coitus-related methods than with the pill or with no method, or perhaps pill users are more likely to be in a phase of the family life cycle where there is more variability in coital frequency. Although the association between the two reports 3 months apart seems quite low, there is not enough of a difference between the consistency of women using the pill and those using other methods to warrant concern about the aggregate comparisons. The somewhat lower consistency of reporting among women using the pill could be due to the higher fre­ quency of a younger group; it is easier to be consistent when the behavior reported is less frequent. Results

We note in Table VII-12 that the mean monthly coital frequency of 9.2 reported by women taking the pill exceeds the average for women using all other methods of contraception (6.6) by 39 percent. The aver­ age for women currently using no method of contraception is 6.1. This is a heterogeneous category, included mainly for completeness, which comprises such diverse groups as women who are pregnant, trying to be­ come pregnant, or risking pregnancy, and those with no need for contra­ ception because of subfecundity. The 39 percent difference among all women under 45 is inflated by the fact that younger women are more likely to use the pill, and they have a higher coital frequency than older women. However, when age is controlled, there is still a considerable (though smaller) difference between women who use the pill and those who use other methods. Although the relative excess is highest among women under 25 (a 29 percent difference), there is still a 20 percent difference among the older women. Our present purposes do not include comparing the coital frequency of women using the pill and those using other methods for all of the po­ tentially relevant subgroups of the population, but we have included in Table VII-12 some summary estimates for women classified by race, education, and religion. In general, there appears to be some group vari­ ation in mean monthly coital frequency—it is slightly higher for white women than for nonwhite women, its correlation with level of education is positive, and it is lower for white Catholic women than for white nonCatholic women. The difference between the rates for women using the pill and those using other methods seems to vary on both the educational and the religious dimension. The higher the education, the more differ-

CHAPTER Vll

ence the pill seems to make, but an age control would be necessary to interpret this finding. For white Catholic women the frequency is 45 per­ cent higher for those using the pill, for white non-Catholic women it is Table VII-12. Mean monthly coital frequency and number of women using the pill, other methods, or no method, by age, race, education, and religion.

Total

Pill

Other Methods

No Method

Percent Pill Over Other Methods

Mean Monthly Coital Frequency Total

6.8

9.2

6.6

6.1

39

Age Under 25 25 - 34 35 - 44

8.8 7.0 5.5

11.0 8.4 6.7

8.5 7.1 5.6

7.5 6.4 5.2

29 18 20

Kace Uhite Nonwhite

6.8 6.5

9.3 8.6

6.7 6.4

6.1 6.0

39 34

Education College High School 4 High School 1-3 Grade School

7.6 7.0 6.8 6.2

9.9 9.3 8.6 8.4

6.9 6.7 6.4 6.4

6.3 6.1 6.2 6.0

43 39 34 31

Religion (Whites) Non-Catholic Catholic

7.1 6.3

9.4 8.7

7.0 6.0

6.2 6.0

34 45

Total

4,600

688

1,871

2,041

Age Under 25 25 - 34 35-44

1,037 1,739 1,824

280 303 105

323 738 810

434 698 909

Race White Nonwhite

3,574 1,026

559 129

1,502 369

1,513 528

Education College High School 4 High School 1-3 Grade School

810 2,003 1,180 605

152 324 171 41

370 879 434 186

288 800 575 378

Religion (Whites) Non-Catholic Catholic

2,524 1,050

449 110

1,042 400

1,033 480

Number of Women

EFFECTS OF PILL VSE

34 percent higher. This difference is due at least in part to the more extensive Catholic reliance on the rhythm method, which is associated with lower coital frequency. What interpretation should be offered for the difference between the coital frequency of women using the pill and women using other meth­ ods? Can the difference be attributed directly to the psychological and physiological properties of the pill? This conclusion does not necessarily follow, since we have no basis for precluding the possibility that couples with higher rates of coital frequency may be more likely than those with lower rates to select the pill because of its unobtrusiveness or, conversely, that women with very low frequencies may regard the various "costs" of the pill as prohibitive for their needs. Since we cannot assign women randomly to experimental and control groups, we will never really be able to satisfy the objection that selection may be involved, because, even if the two groups reported the same frequency prior to the use of the pill by one of them, the argument could still be advanced that the women who adopted the pill had a higher desired frequency than the other women, but repressed it until their adoption of the pill removed the obstacles to its attainment. Nevertheless, even though it would not pro­ vide a final solution to the problem of selection on "desired frequency," the plausibility of our inference about the effect of the pill would be greatly improved if we had a "before and after" report of women who did and those who did not adopt the pill. This opportunity is afforded by data from the Princeton Fertility Study. For the PFS a probability sample of 1,165 married white women who had given birth to their second child six months earlier was interviewed in 1957. The initial sample was confined to seven of the largest metro­ politan areas of the United States, but the women were reinterviewed regardless of subsequent change of residence. In 1960, 905 of the orig­ inal sample were successfully reinterviewed; from 1963-67, 6 to 10 years after the first contact, a total of 814 women were reinterviewed for the third and last time as they approached the end of their reproductive periods. The completion rates were higher than these numbers imply (81 percent and 93 percent for the two interviews, respectively) because women whose marriages were dissolved were classified as ineligible for reinterview. The third interview was spread over 5 years because the oldest women were interviewed first and the youngest women last. The data relevant to this analysis are derived from a question on coital frequency asked in the second interview and repeated in the third: "In recent months, about how frequently have you and your husband had

CHAPTER VIl

intercourse? I mean on the average, of course." Although the wording of the question is different from that used in the NFS, the averages are strikingly similar. The mean coital frequency reported by women in the PFS in the third interview is 7.5 for women using the pill and 6.5 for women using other methods. The corresponding averages from the NFS for similar women (white, aged 30-39, who had at least two children) are 7.6 and 6.4, for users of the pill and other methods respectively. The comparisons of interest are those between the coital frequency rates of 69 women before and after they began using the pill (1960 and 1963-67 respectively) and between the rates at comparable times for the 354 women who used other methods. Since the pill is used more by younger women—who normally have a higher average coital frequency even within this relatively homogeneous group aged 30-39—we calcu­ lated a rate standardized on the age distribution of women using the pill, in order to control any effect of age on the comparison of the two cate­ gories of women. The main observations from Table VII-13 are: (1) a 5 percent in­ crease in mean monthly coital frequency over the 5 to 6 years for women Table VII-13.

Mean monthly coital frequency reported by women interviewed before and after use of the pill and other contraceptive methods.

Mean Monthly Coital Frequency Method Reported in Third Interview

Number of Women

Other Methods

a

(Standardized)8,

1963-67

Percent Change

7.1

7.5

+5

354

7.2

6.5

-10

354

7.4

6.6

-11

69

Pill Other Methods

1960

Standardized, on the age distribution of women using the pill.

Source:

Princeton Fertility Study.

who used the pill; (2) a 10 percent decrease over the same time period for women who used other methods of contraception; (3) no evidence that women with an initially high frequency were attracted to the pill. Given the small number of cases, the change in frequency among women using the pill is not large enough to be statistically significant (using the t test for the significance of differences between means) but the decrease among women using other methods is significant at the 0.001 level. Prior

EFFECTS OF PILL USE

to the use of the pill, the difference between the mean monthly fre­ quencies for the two categories of women is 0.1 in the opposite direction and not statistically significant, but at reinterview the difference is 1.0 in the predicted direction and significant at the 0.02 level. In summary, it seems clear from this analysis that the pill does make a difference in coital frequency and that the 18 to 20 percent difference for women aged 25-44 observed in the NFS (Table VII-12) is quite con­ sistent with the difference observed for women of comparable age in the PFS (a mean monthly frequency of 7.5 for pill users and a theo­ retical rate of 6.3 for the same women if they had not shifted to the pill).10 It is not possible to determine from these data and this type of study the relative importance of social, psychological, or biochemical factors in causing this difference. TIME REQUIRED FOR CONCEPTION

There has been a good deal of speculation and some fragmentary evi­ dence suggesting that the use of oral progesterones facilitates conception after discontinuation; indeed, one of the earliest uses of these drugs was for the promotion of fertility among women of low fecundity, because these compounds regularize the ovulatory cycle and thus make the time of conception more predictable. The data available from the 1965 Na­ tional Fertility Study provide an opportunity to test the hypothesis that women who have used the pill conceive faster after discontinuing use than women who discontinue other methods in order to conceive. Some Problems of Selectivity

To compare the conception times in the two types of experience re­ quires imposing several statistical controls to minimize various possibil­ ities that women who had adopted the pill differ from women using more traditional methods in ways that could distort the comparison. There is the possibility that women using the pill might be more fecund, since they are younger than women using other methods. The best protection against this bias would be to confine the comparisons to women who were of similar age at the time they discontinued contracep­ tion in order to conceive. This control can be reasonably approximated by restricting the analysis to comparisons specific for pregnancy order. At the time of the interruption of contraception for a first pregnancy, 10 This hypothetical rate was estimated by applying the 11 percent decrease for the standardized mean of women using other methods to the mean of 7.1 for women who subsequently adopted the pill.

CHAPTER VlI

women who had used the pill were 21.3 years of age on the average, as compared with a mean age of 21.8 for women who had used other methods. The corresponding mean ages for the second pregnancy are 23.3 and 24.0, respectively. Later planned pregnancy orders do not in­ clude enough women who had used the pill to permit comparison. There is also the more troublesome possibility of memory bias: women who interrupt use of the pill might be apt to remember the event more accurately than women interrupting other methods, since the former is a more radical change in behavior. Whether this memory dif­ ference would operate to shorten or lengthen the time period recalled is not clear. Unfortunately the number of women (7) in our reinterview sample who deliberately interrupted use of the pill in order to become pregnant is too small to permit any comparative analysis. It is clear, how­ ever, that the reliability of such data in general is low enough to warrant concern. For planned first pregnancies in our sample (even restricting the comparison to women who consistently reported that contraception was used and that the pregnancy was planned) only 62 percent con­ sistently reported the number of months required to conceive after inter­ rupting contraception. If the definition of consistency is relaxed to include responses differing by one month, a total of 82 percent replied consistently. The corresponding estimates for planned second preg­ nancies are 59 percent and 76 percent respectively. One control imposed on memory bias was to confine comparisons to recent pregnancy intervals; only pregnancies terminating since June 1, 1961, are included in the present analysis. Such truncation should reduce recall error. This date was selected because the pill was licensed for sale in 1960, and our analysis focuses on time required for conception after contraception is deliberately interrupted. This truncation restricts the intervals of women using other methods to the period since the introduc­ tion of the pill. Even when the above biases are controlled, there remain other differ­ ences between pill users and women who used other methods which lead to an expectation of shorter conception times for women who used the pill—quite apart from any direct physiological effects of the compound itself. Differences in Coital Frequency

The preceding section of this chapter indicates that women who use the pill tend to have higher rates of coital frequency than women using other methods. Insofar as conception time is related to rates of coital

EFFECTS OF PILL USE

frequency, this factor might lead to an expectation of shorter intervals for women who had used the pill. Since we asked only one question about coital frequency, the number of times in the past four weeks, our data do not include any measurement relevant to the conception times under consideration. Therefore, the best we can do is to compare, for women currently trying to become pregnant, the rates of those whose most recent method was the pill with the rates of those who had most recently used some other method. Since the women who had used the pill are much younger, the coital frequency for women interrupting other methods was adjusted to the age distribution of pill users. The mean monthly frequency (in the preceding 4 weeks) for women who had in­ terrupted the pill is 11.7; the mean for women who had interrupted other methods (standardized on the age distribution of the younger women who had used the pill) is 8.6.11 The next obvious question is whether a difference of this magnitude at this level has any influence on the probability of conception. The an­ swer involves many complexities, such as the assumed length of the fertile period, the length of the menstrual cycle, and the distribution of coitus throughout the cycle. A series of estimates of probabilities has been prepared, based on different assumptions about these distributions. From these we infer that an average difference of between 9 and 12 times a month could be expected to reduce the average waiting time from either 2 or 3 months to 1 month if coitus were concentrated in the middle of the month.12 Thus, it is possible that any observed difference in waiting times could simply reflect differences in coital frequency rather than any physiological factor resulting from pill use. At best, however, this is fairly speculative, given all of the assumptions about the relation­ ship of coital frequency to the probability of conception, as well as the assumption that the differences in coital frequency between women cur­ rently using the pill and women using different methods represent other women's past experience. Differences in Knowledge of the Ovulatory Cycle

Considering the fact that women who adopt the pill tend to be better educated than women who adopt or remain with more conventional 11

The unadjusted mean is 7.0. The two values of 2 and 3 months reflect respectively a broad and a peaked fertile period. These estimates are based on a strong assumption that the women are trying to get pregnant (as is the case in our study) and therefore concentrate coitus in the middle of the cycle. See Peter A. Lachenbruch, "Frequency and Timing of Intercourse: Its Relation to the Probability of Conception," Population Studies 21 (1967): 23-31. 12

CHAPTER Vll

methods, any difference in the rates of conception between the two groups might also be influenced by a greater knowledge of the ovulatory cycle among women using the pill. If, for example, a monthly coital fre­ quency of 10 is concentrated around the middle of the cycle, the average conception time would be 1 or 2 months (depending upon the assumed distribution of the fertile period) compared with an average of 3 months if the probability of coitus is equal for any day of the cycle.13 Other dis­ tributions of coitus, such as those that might occur for couples with er­ roneous information about the time of ovulation, could result in much longer waiting times. Our data permit us to determine whether, in fact, women who have used the pill are more knowledgeable about the ovulatory cycle than those who have used other methods by comparing their responses to the question: "If a woman has her period every 28 days, on which day does she have the greatest chance of becoming pregnant, counting from the first day her period begins?" We have used two definitions of the "cor­ rect" answer: a narrow definition including only responses completely enclosed in the range 13-15 days, and a broad definition including any answer falling in or overlapping with the range of 12-16 days. According to the narrow definition, 58 percent of the women who planned their last pregnancy after interrupting the pill gave the correct answer, compared with 48 percent of the women who had used other methods of contra­ ception. The corresponding results for the broad definition are 69 per­ cent and 64 percent. Although the differences are in the expected direction, they do not appear large enough to account for much variation in conception rates between the two groups. We also note that a substantial proportion of the women, whatever the method used, are ignorant of ovulation time. Results

Our interview schedule contained questions designed to permit an approximate separation of each pregnancy interval into components of contraceptive and noncontraceptive exposure for women who had de­ liberately interrupted contraception in order to conceive. The total in­ terval, after subtracting time pregnant and in postpartum, is decomposed into contraceptive exposure and the interval between the time contracep­ tion was interrupted and conception. The waiting time is estimated from the answers to the question: "How many months after you stopped using a method did you become pregnant?" For women who were currently 13

Ibid., p. 30.

EFFECTS OF PILL USE

trying to become pregnant (of whom a somewhat higher proportion are pill users) but who had not succeeded by the time of interview, the wait­ ing-time information derives from the question: "How long ago did you stop using a method?" Even during the limited period of time concerned (June 1961 to Sep­ tember 1965) a much greater proportion of the experience of women who had used conventional methods occurred earlier in the period, whereas the experience of women who interrupted use of the pill was concentrated in the later part of this interval. Consequently, women who interrupted conventional methods have had a much greater opportunity to experience longer waiting times than those who interrupted the pill. This bias can be dealt with by conventional life-table procedures, in which the probability of "surviving" any given month is based only on the experience of women exposed in that month. The cumulative prob­ ability of remaining nonpregnant through a given month is the cumula­ tive product of the preceding monthly probabilities of survival.14 The number of women with more than 6 months of exposure is insufficient to permit reliable estimates of the monthly probability of conception be­ yond the sixth month. Consequently, mean conception times in Table VII-14 are based only on the first 6 months of exposure. The probability of remaining nonpregnant through the twelfth month is presented for general reference, but is based on insufficient cases to be statistically reliable. Two things should be noted from Table VII-14. In the first place, contrary to our expectations, mean conception times are not shorter for women who interrupted the pill than for women who interrupted other methods. Considering the data for all four pregnancy orders combined, 85 percent of the pill users were pregnant by the end of 6 months and their mean waiting time was 2.3 months, whereas for those using other methods 87 percent were pregnant by the end of 6 months, and their mean waiting time was 2.1 months. The differences are in the same direc­ tion when comparisons are made specific for pregnancy order, although the size of the difference between means is larger. Secondly, the distributions of waiting time for pill users and others differ in one particular regard: pill users are less likely to report getting pregnant during the first month of exposure. Again considering the example of the first four pregnancies combined, 46 percent of the women using other methods and 37 percent of the pill users report waiting times 14 R. G. Potter, "Application of Life-Table Techniques to Measurement of Contraceptive Effectiveness," Demography 3 (1966): 297-304.

CHAPTER Vll Table VII-14.

Probability of remaining nonpregnant following the interruption of contraception for women who have conceived or are currently trying to conceive.3

Interval

First Pregnancy Pill Other

Second Pregnancy Pill Other

First and Second Pregnaneies Pill Other

First Through Fourth Pregnancies Pill Other

Month 1

0.63

0.52

0.65

0.57

0.64

0.55

0.63

0.54

2

0.45

0.35

0.53

0.40

0.48

0.38

0.48

0.37

3

0,26

0.26

0.41

0.29

0.33

0.28

0.35

0.27

4

0.19

0.19

0.23

0.25

0.20

0.23

0.24

0.22

5

0.14

0.16

0.14

0.23

0.13

0.20

0.18.

0.19

6

0.09

0.11

0.09

0.16

0.09

0.14

0.15

0.13

12

0.09

0.05

0.00

0.08

0.03

0.07

0.06

0.06

Mean for First 6 lfonths Number of Women

a

2.3 47

2.0 139

2.6 31

2.1 216

2.5 78

2.1 355

2.3 110

2.1 579

Because of the recency of the introduction of the pill, the data are confined to women who interrupted contraception between June 1961 and September 1965·.

of 1 month or less. This difference, if reliable, could result from report­ ing differences between the two groups. In the first place, there may be misreporting of the circumstances of pregnancy. An accidental preg­ nancy reported as "planned" would contribute the shortest possible in­ terval—and the likelihood of such misreporting would be greater for women using less effective methods than for women using the pill. Sec­ ondly, there may be differences in sophistication between the two groups which could be reflected in the interpretation of the question of how long it takes to become pregnant. Taking into account all of the above, it is our considered conclusion that there is no difference in the time required to conceive between women who have used the pill and women who have used other methods of contraception. SUMMARY

About one-third of all women who had used the pill at any time since 1960 had discontinued use, either permanently or temporarily, by the time of the interview, the autumn of 1965. The majority of these women

EFFECTS OF PILL USE

stopped because of what they perceived as unpleasant side effects of the drug, most of which were either undesirable pregnancy-like reactions or problems with the menstrual cycle. Typically, the women who discon­ tinued use of oral contraception had experienced nausea or menstrual breakthroughs and stopped using the pill after one or two cycles. The rates of discontinuation of oral contraception for reasons of side effects, problems of use, and for all reasons were analyzed, following life-table procedures, for women beginning use of the pill in successive periods since 1960. The probability of discontinuation appears to have declined between 1960-62 and later years. Excluding reasons extraneous to the pill itself, such as stopping to have a child, the most recent drop­ out rate was approximately 3 to 4 percent per month over the first 3 months and 1 to 2 percent in subsequent months. The proportion con­ tinuing for at least 1 year was approximately 80 percent. Although women under 30 years of age had a lower dropout rate than older women, the age difference was limited to the first 3 months' experi­ ence, suggesting perhaps that older women may be less patient than younger women with minor side effects. Similarly, women with three or more children had a higher dropout rate than women with fewer chil­ dren, but most of this relationship seemed to be due to the associated age factor. Analysis of the relation of level of education to the probability of dis­ continuation revealed a tendency for women with the least education to experience the highest dropout rates. There was some indication that nonwhite women had a higher dropout rate than white women, but the dif­ ferences were perhaps less than might be expected. Women who discontinued use of oral contraception tended to return to the method they used before the pill. Some slight indication of a tend­ ency to use other new methods, such as foam and the intrauterine device, was also noted. A comparison of the coital frequency of women taking the pill with women using all other methods of contraception reveals a higher fre­ quency for those on the pill. The conclusion is based on evidence both from the 1965 National Fertility Study and from the longitudinal Prince­ ton Fertility Study. In the latter study the same women were interviewed both before and after they adopted the pill, a technique which offers the opportunity to draw a more reliable inference about the pill's effects. The 1965 National Fertility Study also permits an approach to the question of whether women trying to become pregnant after they have interrupted use of the pill succeed in a shorter period of time than wom-

CHAPTER VIl

en interrupting the use of other methods. The analysis of this question is beset with numerous methodological problems, only some of which can be satisfactorily handled. Within the limits imposed by these problems, our conclusion is that there is no difference in the time required for con­ ception between the two groups. In brief, the use of the pill does not, as occasionally asserted, have the effect of making a woman conceive more rapidly after she discontinues it.

CHAPTER VIII

Catholic Conformity Before and After the Papal Encyclical WITH the publication on July 29, 1968, of the Encyclical, Humanae Vitae, Pope Paul ended nearly a decade of lively speculation about the possibility of a revision or reinterpretation of the official Roman Catholic position on birth control. A new period has now commenced which some observers have characterized as one of crisis within the very authority structure of the Church. The vociferous and profound disagreements provoked by the Encyclical both here and abroad have threatened to produce virtually a new schism within the world Church. Behind the public dissent lies the basic and critical question of what the average Catholic has done, now does, and will do in his or her dayto-day decisions about birth control. Was there a significant number of Catholics who, perhaps with the sanction of some parish priest antici­ pating a liberalization of Church dogma, had begun using methods of contraception now reaffirmed as illicit? What proportion of American Catholic couples were conforming to the Church Magisterium's position before the Encyclical? Were those not conforming on the birth control issue simply those who had fallen away from the Church in other ways? And finally, what effect has the Encyclical had on the contraceptive be­ havior of American Catholics? Is anybody listening? These are the primary questions to be answered in this chapter. The chapter is divided into three parts. The first section analyzes the factors associated with Catholic conformity in 1965 and the second sec­ tion describes the trend since 1955.1 The third section assembles evi­ dence from a new follow-up study in 1969 about the effects of the Papal Encyclical on the contraceptive behavior of the Catholic women initially interviewed in 1965. CONFORMITY IN 1965 Between 1955 and 1965 the proportion of Catholic wives using methods of birth control other than rhythm increased steadily. This phe1 These data were presented originally in R. H. Potvin, C. F. Westoff and Ν. B. Ryder, "Factors Affecting Catholic Wives' Conformity to their Church Magisterium's Position on Birth Control," Journal of Marriage and the Family (May, 1968): 263-72; and in C. F. Westoff and Ν. B. Ryder, "Methods of Fertility Control Used in the United States: 1955-1965," in W. T. Liu, ed., Family and Fertility (Notre Dame: University of Notre Dame Press, 1967), pp. 157-70.

CHAPTER Vlll

nomenon occurred despite the fact that most Catholic theologians, prior to 1963, had condemned all forms of birth limitation except periodic or total abstinence.2 Since that date many Catholic theologians have ap­ proved other methods of contraception, but the Magisterium of their Church (the official teaching authority) has not changed its position. In 1964 and again in 1966 Pope Paul affirmed: "The thought and the norm of the Church are not changed; they are those in force in the traditional teaching of the Church."3 And, as the whole world now knows, the Papal Encyclical issued at long last in July, 1968, reaffirmed this traditional position. In effect the 1965 National Fertility Study revealed that a majority (53 percent) of all Catholic wives 18-39 were not conforming or had at some time not conformed to their Church Magisterium's position on birth control, a significant increase from the 38 percent of 5 years before and the 30 percent estimated in 1955—an increase at least in reporting if not in fact. Is this phenomenon diffused throughout the Catholic popu­ lation or is it characteristic of certain subgroups? Have the factors affect­ ing conformity or nonconformity changed between 1955 and 1965? The following analysis based on white women interviewed in 1965, primarily those under 45 years of age, attempts an answer.4 Problems of Definition

Conformity and nonconformity are determined by the wives' reporting having never used any method, having used rhythm only, or having used methods other than rhythm. Such a definition presents certain problems. Fecund women who have "never used" any method are not distinguished from subfecund women in the same category. While the distinction is crucial in assessing the relationship between contraceptive practice and fertility, it is less so in an analysis of Catholic conformity. In fact a good case can be made for including subfecund women who never use con­ traception despite the fact that many are not at risk, because they are in fact conforming. The same case can be made for subfecund women who have used other methods besides rhythm, because in fact they have not 2John Noonan Jr., Contraception (Cambridge, Massachusetts: Harvard Uni­ 1 versity Press, 1965), p. 512. 3 Address given to delegates of the Italian Society of Obstetrics and Gynecology, October 29, 1966; translation (November 7, 1966) by the N.C.W.C. Documentary Service. 4When comparisons with the earlier 1955 and 1960 studies are offered, the base is restricted to women 18-39 years of age. Women under 45 are the base for the cross-sectional analysis in 1965.

BEFORE AND AFTER THE ENCYCLICAL

conformed.5 In addition, although a persuasive argument can be ad­ vanced for trying to develop a measure of conformity that would reflect a decision made by couples who are forced to make such a decision, such refinement is difficult with our data and would imply sacrificing both the level of generality (resulting from use of a base of "all Catholic wom­ en") and the easy comparability with the earlier studies that we now have. One index of conformity that we did develop which approaches greater refinement uses a base of "all Catholic women who have ever used some method of contraception." This index eliminates women who have never used any method of contraception (mainly newly married women) and those who have no need for contraception because of subfecundity. This index was examined in all of the tabulations included here and, although variations in conformity were frequently stronger, the generalizations about relationships did not change at all from those based on the original index. A second and more serious problem for our particular focus is the fact that the category of those who have "ever used other methods" refers to past performance as well as current practice. In other words the classifi­ cation does not distinguish between women who used a method other than rhythm in the past but are now using rhythm (repentant noncon­ formists) and women who are currently using or recently used another method besides rhythm (unrepentant nonconformists). This lack of con­ ceptual precision is especially problematical in the analysis of the relation of conformity to current religious practice. For this reason we seriously considered distinguishing between most recent and past use, but an index of conformity constructed on this basis, when compared with the index based on "all uses," classified 96 percent of the fecund respondents in the same way.8 (Only 30 of 817 women used some other method prior to current or recent use of rhythm.) Given this similarity between the two indices, we decided in favor of the one based on "all use"; again it facilitates comparing our findings with those of the previous studies. Thus, for all Catholic women, conformity to the Magisterium of the Catholic Church means having never used any method of birth control or having used rhythm only. Nonconformity is defined as having used other methods in addition to or instead of rhythm.7 5 In our sample, 25 percent of the Catholic women can be classified as currently subfecund; two-thirds of them have "never used" any method. 6 Women classified as "subfecund" would have to be omitted in an index of current practice, because the duration of time since the last method used varies considerably. 7 Reinterviews three months later revealed that this conformity index is ap­ proximately 87 percent consistent.

CHAPTER VIII

Religiousness and Conformity

In view of the fact that conformity to religious teaching on birth con­ trol can be viewed as an integral part of general adherence to doctrine and involvement in the organized life of the Church, it is appropriate to begin our analysis of factors associated with conformity by examining its relationship with other measures of religiousness. The relationship is presented in Table YIII-1. As would be expected, Catholic women mar­ ried to Catholics report a higher proportion conforming than do Catholic women married to Protestants. The extent of conformity is also associ­ ated with the amount of education received in Catholic schools, the fre­ quency of attending Mass, and especially with the frequency of receiving Table VIII-1. Method of contraception used by Catholic wives under 45, by measures of religiousness.

Method of Contraception Used (%) Rhythm Other Only None Methods

Total Number

Total White Catholic Women

24

25

51

1093

Religion of Husband Protestant Catholic

22 24

16 26

63 49

174 855

Catholic Education None Some All

27 23 16

16 26 50

57 51 34

575 331 177

Mass Attendance Yearly or less Monthly or a few times a year 2-3 time's a month Weekly Weekly or more

28 21 22 25 26

0 4 9 31 42

72 75 70 44 32

65 156 82 737 50

Communion Yearly-or Few times Monthly 2-3 times Weekly or

22 23 24 25 28

5 16 28 38 52

73 62 48 37 20

308 219 220 132 207

33 22 28 26 21 26 23 13

1 3 12 11 23 39 43 60

66 74 60 63 56 35 34 27

55 90 144 178 202 209 161 52

less a year a month more

Index of Religiousness 0-1 Low 2 3 4 5 6 7 8 High

BEFORE AND AFTER THE ENCYCLICAL

• nunion. A summary index which incorporates these latter two variwith six other measures—how often the woman sees a priest or nun a®Out personal problems, whether she was married by a priest, whether religious activities are a part of the daily life of her family, how religiousminded she feels she is, whether she had religious instruction as a child, and whether there were religious activities in the daily life of her family when she was a child—also shows a fairly strong association with con­ formity. We conclude from this analysis that conformity to Church teaching on birth control is very much, although by no means entirely, a part of conformity to Catholic norms in general. Differentials

Previous studies have suggested that the conformity of Catholic women to the Church teaching on birth control varies directly with socio­ economic status.8 In Table VIII-2 distributions of women by the method Table VIII-2,

Method of contraception used by Catholic wives under 45, bymeasures of socioeconomic status.

—.

Husband's Occupation3 and Wife's Education

Method of Contraception Used (%) Ehythm Other None Only Methods

Total White Catholic Women

24

25

51

1093

Husband's Occupation Upper White-collar Lower White-collar Upper Blue-Collar Lower Blue-Collar

21 21 22 29

35 33 20 14

44 46 58 57

298 148 292 288

Wife's Education College 4 College 1-3 High School 4 High School 1-3 Grade School

10 26 20 26 45

57 26 31 11 10

33 48 49 64 45

51 116 553 246 127

White Collar College High School 4 < High School 4

21 21 20 24

34 38 37 18

45 40 43 58

446 117 258 71

Blue Collar College High School 4 < High School 4

25 21 19 32

17 26 25 8

58 54 57 60

580 39 271 270

3

Number

Excludes farm and military occupations.

8 See, for example, P. K. Whelpton, A. A. Campbell, and J. E. Patterson, Fertility atfd Family Planning in the United States (Princeton: Princeton University Press, 1966), p. 281.

CHAPTER Vlll

of contraception they use are presented for different levels of education and the husband's occupation. In general the proposition still appears to hold in 1965: conformity is more common in the white-collar classes and among the better educated women. As the two bottom panels indicate, occupation and education seem to operate with some degree of inde­ pendence of one another. Whether or not these differences have nar­ rowed over time (since 1955) will be assessed later. Since religious practice among Catholics is also related directly to socioeconomic status,9 the question arises whether or not the socioeco­ nomic factor in conformity is reducible to this religious factor. Holding constant a crude division of the reported frequency of receiving Com­ munion and varying conformity by occupation and education (Table VIII-3), we find that among those who might be called the less active Catholics (who receive Communion less than monthly) the higher status women conform the least, while among the more active Catholics higher socioeconomic status means greater conformity. In one Table VIII-3.

Method of contraception used by Catholic wives under 45, by socioeconomic status, and religiousness.

Husband's Occupation, Wife's Education and Frequency of Receiving Communion

Method of Contraception Used (%) Ehythm Other None Only Methods

Number

Total Hhite Catholic Women

24

25

51

1093

Less than Manthly White Collar Blue Collar

18 24

11 9

71 67

1-62 334

College High School 4 < High School 4

16 16 32

5 13 6

79 71 62

55 230 242

23 29

48 28

29 43

280 243

23 22 34

50 44 19

26 34 47

111 320 128

At least Monthly White Collar Blue Collar College High School 4 < High School 4

a

HStcludes farm and military occupations.

9 See, for example, George A. Kelly, Catholics and the Practice of the Faith (Washington, D.C.: Catholic University Press, 1946); Joseph H. Fichter, Social Relations in the Urban Parish (Chicago: University of Chicago Press, 1954); Herve Carrier, The Sociology of Religious Belonging (New York: Herder and Herder, 1965).

BEFORE AND AFTER THE ENCYCLICAL

sense, the association between socioeconomic status and contraceptive practice among the less active Catholics is similar to that among nonCatholics.10 Interpretation of this reversal is complicated by the circum­ stance that college-educated Catholics receiving Communion frequently are more likely to have gone to Catholic colleges, while college-educated Catholics who receive Communion infrequently are more likely to have gone to non-Catholic colleges.11 The covariation of conformity with education and frequency of receiv­ ing Communion may be considered in another way. The proportion of Catholic women with college educations who have a high Communion frequency is 67 percent; for those with less education it is only 35 per­ cent. This educational difference in Communion frequency would be sufficient by itself to explain the variations in conformity. The independ­ ent effect of education is seen in the comparison of proportions conform­ ing for the higher and lower Communion-frequency categories within each education subgroup. The higher the level of education, the stronger the relationship between Communion frequency and contraceptive conformity. The same kind of analysis holds for occupational differences. The variation between white-collar and blue-collar conformity can be ex­ plained by the difference in frequency of receiving Communion. Further­ more, the relationship between Communion frequency and contraceptive conformity is much stronger within the white-collar subgroup than within the blue-collar subgroup. Presumably this relationship is interdependent with that between contraceptive conformity and education. Thus the socioeconomic variables play an interesting but essentially secondary role in determining the extent of contraceptive conformity. The interrelations of conformity with religiousness and socioeconomic status invite consideration of the role of ethnic origin. Earlier work has shown that Catholics of Irish origin, having migrated to this country ear­ lier than many other groups, tend to be more concentrated in the whitecollar classes and, representing a more orthodox variant of Catholicism, tend to be more involved in religious activities and more committed to traditional religious beliefs.12 At the other end of the scale are Catholics of Italian origin who, having migrated more recently, occupy the lower 10

See Chapter V. See, for example, C. F. Westoff and R. H. Potvin, College Women and Fer­ tility Values (Princeton: Princeton University Press, 1967). 12See, for example, C. F. WestofF, R. G. Potter, and P. C. Sagi, The Third Child (Princeton: Princeton University Press, 1963), p. 239; Westoff and Potvin, op. cit., pp. 146-152. 11

CHAPTER Vlll

rungs of the socioeconomic ladder and represent less formal versions of Catholicism. The expected polarity of women of Irish and Italian backgrounds is clearly supported by the data presented in Table VIII-4. Whereas 41 percent of the women of Irish origin report ever having used methods of contraception other than the rhythm method—the lowest proportion of all groups (but equal to the French,13 who are mostly French CanaTable VIII-4. Ifethod of contraception used by Catholic wives under 45, by ethnic background of their parents.

Method of Contraception TJsed (¾)

Ethnic Background

None

Total White Catholic Women3 Father same nationality Father other nationality

Number

Ehythm Only

Other Methods

24 27 17

25 24 27

51 49 56

1025 685 340

Mother Irish Father same Father other

22 22 22

37 41 35

41 37 43

167 41 126

Mother French Father same Father other

35 61 19

25 12 32

41 27 49

113 41 72

Mother German Father same Father other

23 28 20

31 31 31

46 41 50

185 68 117

Mother Slavic Father same Father other

21 21 22

23 25 15

56 54 63

131 104 27

Mother British Father same Father other

17 19 16

28 28

56 53 57

145 43

Mother latin-American or Spanish Father same Father other

34 38

10

22

16

56 54 63

145 113 32

Mother Italian Father same Father other

20

18

13 24

62 68

168

19 21

27

8

55

102

88 80

Excludes women with mothers of nationality background other than those listed here. 13 The high conformity of women of French origin results from the high pro­ portion of women who have never used contraception rather than from any par­ ticular reliance on the rhythm method.

BEFORE AND AFTER THE ENCYCLICAL

dians)—62 percent of the women of Italian background are in this category. Women of Latin-American or Spanish background also show a low proportion conforming, and an especially low proportion using the rhythm method. Except among the Italian group, conformity appears to be greater when both of the woman's parents are of the same national background. This may be due partly to the fact that women who mar­ ried outside their ethnic group were more likely to have married nonCatholics. Whether or not the ethnic differential in general can be reduced to differences in religious practice is the focus of the analysis presented in Table VIII-5. When the relationship between the woman's mother's ethnic background and conformity is examined, within a control for the frequency of receiving Communion, some ethnic differences disappear and others are reduced. Among the less active Catholics, women of Irish and French backgrounds still appear to conform the most, but the LatinAmerican and Spanish group now falls in the same category, particu­ larly because of the high proportion of nonusers (rather than users of rhythm). Within the group of more active Catholics, the French and Irish still show the lowest reliance on other methods, but the differences Table VIII-5.

Method of contraception used by Catholic wives under 45, by frequency of receiving communion and ethnic background of woman's mother.

Communion and Mother's Ethnic Background

Method of Contraception Used (%) Othej: Ehythm Methods None Only

Number

Total White Catholic Women

24

25

51

1093

Less than Monthly Irish French German Slavic British Latin-American or Spanish Italian

20 30 26 16 13 38 18

20 11 8 13 5 2 7

60 59 66 71 82 60 75

56 53 61 61 61 98 113

At Least Monthly Irish French German Slavic British Latin-American or Spanish Italian

24 40 22 27 20 28 25

•44 38 42 32 •43 22 40

•32 22 36 41 37 50 35

111 59 121 68 82 47 55

CHAPTER VIII

are not great. In general the variance of contraceptive conformity among Catholic women classified by national background is reduced when reli­ gious practice is controlled, and it is quite evident that religious practice is by far the more important variable. Marriage History Factors

Conformity to a norm prohibiting the use of contraception other than the rhythm method would seem to become more difficult in the later years of marriage when the problem is to control the size of the family rather than the timing of births. Women who marry younger will have had a longer period of exposure to the risk of pregnancy and therefore may be more likely to deviate from religious norms. Examination of the two variables, marriage duration and age at marriage (see two top panels of Table VIII-6), reveals no particular pattern of association between Table VlXI-6.

Method of contraception used by Catholic wives under 45, by marriage duration and age at marriage.

Duration and Age

Method of Contraception Used (%) Rhythm Other None Only Methods

Number

Total White Catholic Women

24

25

51

1093

Years Married Less than 5 5-9 10-14 15-19 20 or more

31 19 20 23 31

25 24 32 23 16

44 58 48 53 53

227 221 235 256 154

Age at Marriage Under 20 20 and older

23 25

16 31

61 44

464 629

Married Less than 5 Years Married under 20 Married at 20 or older

27 34

17 31

56 35

100 127

Married 5-9 Years Married under 20 Married at 20 or older

13 22

17 28

70 50

92 129

Married 10-14 Years Married under 20 Married at 20 or older

18 .20

16 41

66 39

87 148

Married 15 Years or More Married under 20 Married at 20 or older

27 26

15 26

58 48

185 225

BEFORE AND AFTER THE ENCYCLICAL

conformity and the number of years married but does indicate a sub­ stantial difference by age at marriage. Among women married before age 20, 61 percent report having used other methods, compared with 44 per­ cent of women who married at an older age. And this relationship re­ flects a distinct difference in the use of the rhythm method (rather than in use of no method). Since duration of marriage and age at marriage are related, and since the longer a couple is married, the more chance they have to deviate from the norm, the relationship between age at marriage and conformity should be examined with some control on the length of time married. This control however (see bottom half of Table VIII-6) does not diminish the relationship, and the generalization that women who marry young conform less remains intact. We know from the current study as well as from others that age at marriage among Catholic women is directly associated with religious­ ness—only 23 percent of women married before age 20 receive Com­ munion more than once a month, compared with 37 percent of those married at 20 or older—from which we might infer that those women who married younger might also be less likely to conform to traditional rules about the practice of birth control. We also know that other factors, such as amount of education and ethnic background, are jointly related both to conformity and to age at marriage. It came as something of a surprise therefore to discover (in data not presented here) that none of these variables "explained" the relationship. In brief, women married before age 20 conform less than those married at a later age regard­ less of length of time married, religiousness, education, or national background. A measure of the relationship between conformity to Catholic teach­ ing and the use of contraception for timing or controlling the number of births which is more direct than that afforded by marriage duration is available in the form of responses to our question about whether or not the woman wanted to have any more children.14 The tabulation of our conformity index by this simple dichotomy (top part of Table VIII-7) reveals a sharp difference in the expected direction, although the greater nonconformity among women who do not wish any additional children is due mainly to a difference in the proportions who have never used any method rather than in the use of rhythm. Differentiating women on the basis of whether or not they want additional children does appear to in14Women with no children were not asked this question but were assigned to the appropriate category on the basis of a subsequent question on whether the number of children they wanted was greater than zero.

CHAPTER VIIl

fluence the relationship between age at marriage and conformity. Al­ though in both groups women married before age 20 conform less than those married at an older age, the difference is much greater among women who have not reached the desired family size. In other words, after the achievement of desired family size, age at marriage becomes far less significant for conformity. Table VIII-7.

Method of contraception used by Catholic wives under 45, by fertility intentions and age at marriage.

Method of Contraception Used (X) Other Rhythm None Methods Only

Number

Total White Catholic Womena

24

25

51

1093

Desired Fertility Incomplete Married under 20 Married at 20 or older

33 28 36

26 17 32

41 55 32

511 196 315

Desired Fertility Complete Married under.20 Married at 20 or older

16 18 14

23 15 30

61 67 56

570 266 304

a

Includes 12 women not classifiable on whether they have had all the children they want.

It is obvious that we have not really "explained" the significance of the age at which Catholic women marry for their conformity to the Church's teaching. This is partly because there are so many variables associated with age at marriage. We have already mentioned differences by reli­ giousness, education, and ethnic background, and their significance for the time available to achieve the desired family size. And there may be other factors that we can only enumerate, such as the possibility that couples not prepared to risk early childbearing with the rhythm method may postpone marriage, or that couples who are reluctant to abstain periodically from sexual relations may be reluctant to postpone marriage. (Not only are they more likely to have a premarital pregnancy which would force early marriage, but they may also be reluctant to use the rhythm method.) These are just a few of the possible selective factors that may be involved in the relationship between age at marriage and conformity to the Catholic teaching on birth control, a subject that un­ doubtedly deserves further investigation.

BEFORE AND AFTER THE ENCYCLICAL

TRENDS IN CONFORMITY, 1955-1965 As noted at the beginning of this chapter, the recent trend among Catholics is toward more deviation from the traditional position of the Church on birth control. From comparable data collected in 1955 and 1960, we are able to construct a table showing trends by cohort. In Table VIII-8, the percentages in each of the top diagonals are from the 1955 Study; in the middle diagonals, from the 1960 Study; and in the lower diagonals, from the current 1965 Study. These data can be analyzed in TableVHI-8.

Cohort

Method of contraception used by Catholic wives in 1955, 1960, and 1965, by age and birth cohort.9

20-24

.25-29

,30-34

35-39

40-44

45-49

Percent Who Have Used No Method

1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

45 43 32 21

40 25

37

28

21

23

30 30

35

21

22

Percent Who Have TJsed Rhythm Only

1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

27 25

33 39 31

23 35

27 26 22

25 24

13

22

26 Percent Who Have Used Other Methods

1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

30 43 53

30 40 48

37 40 56

28 46 55

45 46

52

Number of Women 1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

a

212

220 128 112

176 176

174 242

209 192

178 241

206

189

Top diagonal 1955, middle diagonal 1960, lower diagonal 1965.

197

89

CHAPTER VIII

three dimensions: (1) vertically, the numbers permit assessment of the trend across cohorts at the same age; (2) horizontally, the percentages show changes associated with time as the cohort ages; (3) diagonally, cross-sectional comparisons of the relation of conformity to age can be made at three different time periods. The most relevant comparison for our interest in time trends is the first or vertical reading. The third panel—the percent who have used other methods—provides the most direct picture. It is quite clear that the increase in nonconformity is occurring across all cohorts at each age group. The source of this change is evident from making the same com­ parisons in the two top panels. The proportions using rhythm do not vary much or systematically across cohorts, but the proportions who have never used any method show consistent declines up to age 40-44. This complementarity of decreases in the proportions who have never used any contraception and increases in the proportions using methods other than rhythm, is evident when the table is read horizontally and the history of each cohort followed as it ages. (The rise in some of the pro­ portions never using, among the older ages of the earlier cohorts, may be due to some differences in procedure in the 1960 Study, to sampling error, or to problems of the reliability of recall among older women.) In general it appears that between 1955 and 1965 the proportion of Catho­ lic women conforming to their Church Magisterium's position on birth control has been decreasing and that deviation from this norm has been occurring at progressively earlier ages over this period. Of the task originally set, one question remains to be investigated. Have various subgroups of the Catholic population participated equally in the trend (between 1955 and 1965) away from conformity? Have the factors affecting conformity changed over this period? Several interesting observations based on the three surveys can be made from the data (for Catholic women 18-39) assembled in Table VIII-9. All socioeconomic subgroups and the more active as well as less active Catholics have par­ ticipated across the decade in the shift away from nonuse of contracep­ tion and toward the use of methods other than rhythm. In the more re­ cent period—between 1960 and 1965—a larger proportion of the more religious Catholics have shifted to the use of other methods, but of course they started from a much lower point in 1960. (The classification by husband's occupation does not produce any outstanding differential time trends.) The most interesting pattern of change, however, is that revealed in the classification by the woman's education. Between 1955 and 1960 the

32 25

40 35 27 21

33 13

23 21

41 32 20 16

27

19 18 19 29

18 18 15 45

22

37 10

44 38 30 21

49 37 20 16

31

33 5

35 36 21 12

38 31 10 10

25

22 47

26 23 32 34

20 30 38 26

30

31 65

34 39 41 50

24 38 57 43

38

44 74

46 46 60 59

44 52 65 45

53

Method of Contraception Used (%>) Rhythm Only Other Methods 1965 1955 1960 1965 1955 1960 1965

533 254

137 82 214 263

73 396 208 110

787

1955

525 143

165 95 143 219

79 340 168 81

668

1960

Number

605 238

232 116 221 218

136 438 177 94

845

1965

Regular means "regularly" in the 1955 and "once a week" in the 1960 survey in answer to questions on the frequency of attendance at religious services. In the 1965 survey the category means "once a week or more" in answer to a question on attendance at Mass.

45 40

34 42 41 45

Husband's Occupation Upper white Lower white Upper blue Lower blue

Regular Less frequent

27 24 23 41

38 39 40 58

Education of Wife College High School 4 High School 1-3 Grade School

21 22 29 29

30

1955

None 1960

Method of contraception used by Catholic wives 18-39, by education, occupation, and frequency of attending Mass: 1955, 1960, and 1965.

All White Catholic Women 43

Characteristic

Table VIII-9.

CHAPTER VllI

increase in nonconformity was inversely associated with her level of edu­ cation, but between 1960 and 1965 the differential was completely re­ versed, with education being directly related to the increase in noncon­ formity. Although such comparisons over time are complicated by the changing distributions of the population in the different educational categories (having gone to college or never having gone to high school has a different meaning in 1965 than in 1955), it seems likely that the disproportionately greater break with traditional Church teaching by the more educated (and the more devout) between 1960 and 1965 is re­ lated both to the advent of the pill and perhaps to the greater exposure of the more educated women to all of the publicity about the delibera­ tions within the Church. There is an understandable temptation to infer that the trend toward nonconformity among Catholic women is indeed a response to the delib­ erations of Church officials about fertility regulation, reflecting confusion in the public mind about the possibility of change in the official position. Although this may be part of the explanation of the acceleration in the decline of conforming behavior between 1960 and 1965, it does not seem satisfactory as a complete explanation of the change, because the change represents a continuation of a trend already observed between 1955 and 1960. Furthermore, the acceptance of oral contraception, which is prin­ cipally responsible for the increase in nonconforming behavior among Catholics between 1960 and 1965, is paralleled by an even greater adop­ tion of the pill by non-Catholics. One of the most significant findings of the 1965 National Fertility Study has been the increase in the use by American married couples of fertility regulation in general and oral con­ traception in particular, a proposition that holds for Catholics and nonCatholics alike. CONFORMITY AFTER THE ENCYCLICAL

With a dateline Rome, July 28, 1969, the New York Times carried the following item: Humanity should be grateful to Pope Paul for his birth-control ban, which saves mankind from itself and especially from the dehumaniz­ ing effect of unbridled sexuality, L'Osservatore Romano, the Vatican newspaper, said in an article today marking the first anniversary of the Pope's encyclical. With the approach of this anniversary we felt that it would be emi­ nently desirable to evaluate what effect, if any, the papal encyclical has

BEFORE AND AFTER THE ENCYCLICAL

had on the contraceptive behavior and conformity of American Catho­ lics. According to a Gallup Poll, more than half (54 percent) of Ameri­ can Catholics opposed the Pope's encyclical on birth control and only 28 percent favored it. (The remaining 18 percent had no opinion.) And we knew from our own data (Table VIII-I) that as of 1965 the majority of American Catholic married women were using methods of birth con­ trol proscribed by the Church. We had the splendid opportunity offered by the 1965 National Fertility Study to observe changes in behavior be­ tween 1965 and 1969 by recontacting those Catholic women interviewed in 1965 and questioning them on their current practices as well as their attitudes toward the Pope's position. Accordingly, we attempted to reinterview by telephone all appropriate Catholic women in the spring of 1969 to determine whether "anyone was listening." The Sample

We aimed at reinterviewing all Catholic women who had been clas­ sified in 1965 as white, under 40 years of age, and fecund. The racial criterion was imposed because the small number of nonwhite Catholics would have made their separate treatment in the analysis difficult; the age and fecundity restrictions were desirable since we were primarily concerned with changes in the methods of fertility control employed and older women and those who are sterile are obviously not appropriate subjects for such inquiry. There were 634 Catholic women interviewed in 1965 who were eligi­ ble according to these criteria to be reinterviewed in 1969. We decided to use the telephone, after conducting a small pretest to determine its feasibility, because it was more economical than a personal interview and because we were interested only in obtaining a limited amount of in­ formation (requiring no more than 15 minutes to secure). We had in­ formed our respondents in 1965 that we might be recontacting them and secured their telephone numbers and the names and addresses of persons likely to know where they would be if they were to move. A total of 1,293 telephone calls were made by female interviewers at the Institute of Survey Research at Temple University where the field work was conducted. These calls produced completed or terminated in­ terviews with 508 of the 634 women for a completion rate of 80 percent. This result was somewhat disappointing; evidently the 3.5-year interval was too long to permit re-establishing contact with a higher percentage. The bulk of the nonrespondents—81 of 126—simply couldn't be located despite several kinds of effort. Remarkably, only 11 women refused to

CHAPTER Vlll

be reinterviewed. The remaining 34 women fall into miscellaneous cate­ gories including no telephone listed, moved out of the country, and so forth. Of the 508 successfully recontacted, 35 did not complete interviews for one reason or another. Since enough information to be useful was not collected from these 35 women, we have combined them with the 126 nonrespondents for analysis of the differences between the women who were and those who were not reinterviewed successfully. Sample Bias

On the basis of information collected in the original 1965 interview, we have made comparisons for certain characteristics of the 473 women reinterviewed successfully and the 161 women never reinterviewed or with whom interviews were not completed. Under the headings of vari­ ous background characteristics, religious characteristics, fertility, and family planning statuses we have assembled various comparisons of the two groups (Table VIII-10). The analysis provides both an estimate (based on the chi-squared test) of the statistical significance of differ­ ences between the two groups and abbreviated statistics designed to illus­ trate the differences and describe the sample in general. In terms of background characteristics in 1965, those women success­ fully reinterviewed in 1969 were slightly older (by one year), concen­ trated more in the Northeast and Midwest than in the South and Far West, better educated, married to better educated husbands more of whom were employed in white-collar occupations and earned more money, and had a much lower proportion of parents on both sides descended from Latin-American (Mexican and Puerto Rican) back­ grounds. The two groups did not differ in age at marriage, marriage dura­ tion, or wife's work history. In brief, the reinterviewed respondents came from a higher socioeconomic background. The successfully reinterviewed also seem to be more religious. By comparison with those not reinterviewed, a higher proportion were mar­ ried by a priest, attended Mass and received Communion more fre­ quently, and scored higher on summary indices of religiousness. They did not differ significantly in the proportion of mixed marriages or in propor­ tions living in Catholic neighborhoods or having Catholic friends. No sig­ nificant difference separates the two groups in measures of past or fu­ ture fertility. The average numbers of children born, desired, or ex­ pected in the future were very similar if not identical. The final area investigated—one that is particularly relevant to the

Table VXII-1Q. Differences for selected characteristics in 1965 between respondents and nonrespondents to the telephone reinterview.

Successfully Reinterviewed

Not Reinterviewed

634 28.4 20.6

473' 28.6 20.7

161 27.6 20.2

8.4

8.5

8.0

Total Sample Background Total Number Mean Age of Wife Mean Age at First Marriage Mean Number of Years Since First Marriage

P*

.05 HS HS

Region of Residence (¾) Northeast Midwest South Far West

38 33 13 15

40 36 12 13

35 25 17 22

,01 .01 .01 .01

Education (%) Wife H.S. 4+ Husband H.S. 4+

69 68

76 75

49 52

.001 .001

Husband White-Collar· (%)

42

45

31

.01

ι 6,870

7,020

6,440

.05

Wife Currently Working (%)

26

28

25

HS

Parents Both Mexican or Puerto Rican (%) Wife1s Husbandfs

13 12

8 8

27 24

.001 .001

19 84

19 87

19 73

NS .01

71

75

59

.001

52 4.8 4.6

64 4.9 4.8

45 4.4 4.1

.02 .05 .02

61

62

60

NS

77

78

74

HS

3.2 2.8 2.7 3.7 52

3.2 2.8 2.7 3.7 52

3.1 2.7 2.6 3.7 51

HS NS NS NS NS

1.3 4.0

1.3 4.0

1.3 3.9

NS HS

Mean Annual Income of Husband (dollars)

Religious Characteristics Percent Non-Catholic Husbands Percent Married by a Priest Percent Attend Mass at Least Once Weekly Percent Receive Communion Once Monthly Index of Religiousness (Wife) Index of Religiousness (Husband) Percent with Half or More Catholic Neighbors Percent with Half or More Catholic Friends Fertility Mean Number of Pregnancies Mean Number of Live Births Mean Number of Children Mean Number of Children' Desired Percent Intend More Children Additional Number of Children Expected Total Number of Children Expected Contraception (%) In Favor in General Using None or Rhythm Only Ever Used Pill Mean Score on Attitudes Toward Abortion Fertility Planning (¾) Completely Planned Excess Fertility

55 54 27 3.5

16 15

55 53 29 3.6

15 15

55 56 22 3.4

19 .14

NS NS .05 HS

HS HS

£ is the probability, according to the chi-squared test, that the differences observed between two distributions could have occurred by chance. NS(not significant) refers to ρ >.05.

CHAPTER VUl

study—is contraception. The two groups were equally in favor of con­ traception and held about the same attitude toward abortion. There was no difference in the proportions conforming (by using no method or rhythm only) to Church doctrine on birth control. There is, among those reinterviewed, a somewhat greater proportion who had used the pill. And finally, the two groups were the same in the extent to which they had planned fertility successfully. In summary, the 473 Catholic women successfully reinterviewed, in comparison with the 161 nonrespondents, came from a higher socioeco­ nomic status and were more religious, but their fertility and contracep­ tive behavior were similar. Although it is difficult to comprehend fully the significance of these biases for the reliability of our findings about changes in behavior for married Catholic women as a whole, it seems to us that the bias toward greater education and closer involvement with the Catholic Church suggests that we are generalizing about a group more likely to be responsive to the Encyclical. Together with the absence of any disturbing differences in fertility or contraceptive behavior, this would seem to imply that we should expect a higher level of conformity after the Encyclical; if none is observed, we can be reasonably assured that it does not result from a biased sample. Content of the Interview

The respondent reached on the telephone received the following introduction: Hello, Mrs. . I am Mrs. , I am calling for Princeton University. You may remember that we interviewed you about three years ago in our study of American family life. We said then that we hoped to get in touch with you again. This will only take a few minutes of your time. If the respondent was too busy to talk then, the interviewer tried to ar­ range a more convenient time to call back. In order to avoid bias, the respondent was never told that the reinterviews were exclusively with Catholics, nor was the Papal Encyclical even mentioned until late in the interview, after all of the important questions about contraceptive practice had been asked. The interviewer began by recalling the respondent's parity in 1965 and then asked about her pregnancy history and contraceptive practice since 1965. In order to avoid the awkwardness of respondents' having to use the names of different contraceptive methods—especially considering the possibility

BEFORE AND AFTER THE ENCYCLICAL

that their conversation could be overheard by other family members— we phrased all such questions so that they could be answered either affirmatively or negatively, limiting the questions to use of the rhythm method, the pill, and "other methods." Although such a procedure does not produce data for the entire battery of methods, it may in fact yield a more complete and reliable report of contraceptive use than the con­ ventional approach. The interviewer concluded with an open-ended question designed to elicit the woman's attitude toward the Encyclical, and a short series of questions on how often she and her husband attend Mass and receive Communion. Results

The primary question we are trying to answer is whether or not the Papal Encyclical has had any effect on the methods of contraception used by Catholics in the United States. More precisely we wish to deter­ mine whether there has been any change in the proportion of women using methods not approved by the Catholic Church, the direction and magnitude of that change, and whether such change could be reasonably attributed to the pronouncement of July 29, 1968. We have approached the problem in several ways. The first is to re­ gard the 1969 survey as simply another cross-section of American Catholic married women with which we can compare the 1965 survey to determine whether there has been any change in the aggregate. An­ other approach is to examine changes in the contraceptive behavior of the same women between 1965 and 1969. A basic problem in the analy­ sis approached from any direction is to devise an estimate of what con­ traceptive practices Catholics would have been employing if the Encycli­ cal had not been issued and—perhaps even more problematic—the pat­ tern of methods used if the Pope's decision had been to liberalize the position of the Church. Suppose, for example, that we find no change in the proportion of women conforming to Church authority in the matter of birth control. Would we interpret this to mean that the Encyclical has had no effect? Or should we infer that it arrested a process of attrition, that it stopped a trend toward nonconformity? These are difficult prob­ lems indeed, and all we can hope to do is to provide some approxima­ tions and try to make reasonable inferences. One way of estimating what might have happened is to measure the rate of change in Catholic conformity in the matter of birth control. As we saw earlier in this chapter, the trend toward nonconformity ac­ celerated sharply between 1960 and 1965. We begin then with an analy-

CHAPTER VIlI

sis of the methods of contraception used by Catholics, as reported in 1955, 1960, 1965, and 1969 (Table VIII-11). The only measure com­ parable across all studies is that of "methods ever used" which, as we have seen in previous analyses, has many limitations and is even more limited in this analysis. It does, however, offer a simple measure and a broad time perspective. Since the 1969 Study cannot include any women married less than 3.5 years, the distributions from each survey presented in Table VIII-11 are based on women married at least that amount of time. The samples are further confined because of the restriction, im­ posed on the 1969 sample, to women classified as fecund.15 Although these restrictions on marriage duration and fecundity prevent us from using the results for estimating general population parameters, they should not appreciably affect our present objective of assessing change. Finally, Table VIII-11 also divides Catholic women by the frequency with which they attend church,16 a separation made both for its substan­ tive interest as well as its methodological advantages (control of the bias toward religiousness characterizing the successfully reinterviewed sample in 1969). Examining first the overall trend, we see that the proportion of Catho­ lic women who have ever used what we have classified as "other meth­ ods"—that is methods other than rhythm exclusively—has increased over the nearly 15-year span from 49 percent in 1955 to 76 percent in 1969. The amount of annual change was approximately the same be­ tween 1965 and 1969 as between 1960 and 1965. Perhaps a more sensi­ tive measure of conformity is the proportion of the women who have ever used a method who rely on rhythm exclusively. These estimates appear in the lower panel of Table VIII-11. Here we get a picture of no change between 1955 and 1960 but a sharp decline in the use of rhythm between 1960 and 1965 which continues between 1965 and 1969. Most of the trend toward nonconformity in recent years has been caused by the changing behavior of the more religious segment of the Catholic population, which is the considerable majority by our crude definition (based on frequency of church attendance). 15 In addition, the 1969 sample are women who were married to the same hus­ band at the time of the 1965 survey. This restriction does introduce some slight noncomparability with the 1955, 1960, and 1965 samples, since they were not confined to women with such marital status specifications 3.5 years prior to those dates. 16 The questions and coding in 1955 and 1960 differed from those of the later surveys. In the former they were phrased in terms of attendance at "religious services"; in the latter "attendance at Mass" was used. The 1955 response category "regularly" is here coded as "more than once a month."

100 397

100

292

100

280

Percent Total

N u m b e r of- C o u p l e s

69

31

Methods

43

57

42

58

Rhythm Only

424

318

330

Other

Couples

Number of

100

100

Percent Total

65

52

6 29

40

8

Total 1960 1965

100

49

Methods

Other

36

15

Only

Method

1955

320

100

78

22

8

268

100

46

46

337

100

60

35

5

4

70

26

264

100

Couples

217

100

51

49

248

100

50

50 37

320

100

63

254

100

73

27

least

63

100

84

16

44

100

93

7

50

75

85

77

100

96

4

87

100

82 100

4

11 6

12

100

71

13

16

66

100

97

3

67

100

96

3

1

Once a M o n t h or Less 1960 1969 1955 1965

C o u p le s W h o H a v e U s e d a M e t h o d

255

100

43

42

15

Percent of All

More Than Once a Month 1960 1969 1955 1965

Percent of

331

100

76

21

3

1969

M e t h o d s of c o n t r a c e p t i o n ever u s e d b y w h i t e Catholic fecund w o m e n u n d e r 4 0 , m a r r i e d a t 4 y e a r s , b y f r e q u e n c y of church a t t e n d a n c e , in 1 9 5 5 , 1 9 6 0 , 1 9 6 5 , a n d 1 9 6 9 .

Rhythm

No

Table VIII-11,

CHAPTER Vlll The data in Table VIII-Il are based on women under 40 married at least 4 years. One consequence is that a large fraction of the respondents being compared in 1965 and 1969 are the same women, whereas the 1955 and 1960 studies are samples of different women. This means that the proportion of women classified in 1969 as never having used any method or as having used rhythm only cannot be higher than in 1965. To minimize this problem we have prepared Table VIII-12, showing basically the same calculations as those in Table VIII-Il but showing Table1VIII-12. Proportions of fecund Catholic women under 40, who have never used any method'of contraception or who have used only the rhythm method, by marriage duration in 1955, 1960, 1965, and 1969.

Karriage Duration

1955

Percent 1960 1965

5-9 10-14 15+

51 48 54

52 48 40

36 39 29

24 22 24

5-9 10-14 15+

42 38 46

47 43 37

29 37 25

20 21 22

1969

1955

Number of Couples 1969 1960 1965

Never or Ehythm Only 190 93 50

149 114 58

166 146 113

135 101 100

135 103 54

150 142 106

129 100 97

Ehythm Only 162 78 43

them separately for the different marriage-duration categories. Thus a comparison of women married, for example, 5-9 years in 1955, 1960, 1965, and 1969 will involve different women at each period (except for some overlap because the interval between 1965 and 1969 is less than five years). The marriage-duration categories 5-9 and 10-14 offer the cleanest comparisons; they disclose a pattern similar to that described above—a decline in conformity from 1960 through 1965 to 1969. Recent Changes in Practice There is another limitation of the "ever used" measure. A woman who had used the pill, let us say, before the Encyclical and who had changed to rhythm after July 1968 would still be classified in the "other methods" category in 1969 because she had not relied exclusively on rhythm dur­ ing her previous marital history. We have two additional choices—a classification by the method currently being used or one by the method most recently used. The latter was selected because it permits classifica-

BEFORE AND AFTER THE ENCYCLICAL

tion of not only current users but also those women who are temporarily not using any method because they are pregnant or trying to become pregnant. Since the bulk of women are using some method, the classifica­ tion is very similar to that for method currently used. The first column of Table YIII-13 was prepared in order to estimate what the distribution of methods used 4 years after the 1965 Study might have been as a result of the fact that the women had been married several Table VIII-13.

Method of contraception used most recently by fecund Catholic women under 44, married at least 4 years,a in 1965 and 1969.

Method of Contraception Used Most Recently

No Method

Percent of Total 1969 1965

7

4

Ehythm Only

35

32

No Method or Rhythm Only

42

36

Pill

14

37

Other Methods

44

27

Percent Total

100

100

Number Total

355

386

Percent of Users 1965 1969

38

34

332

371

Both tabulations are further restricted to women who were still married to the same husbands in 1969 and who successfully completed the 1969 interview.

years longer and were several years older. In effect, this is the synthetic cohort technique applied to adjust for the difference in marriage duration and age. This column shows the distribution of methods used most re­ cently as of 1965 and is restricted to fecund women married at least 4 years17 who were less than 44 years of age in 1965. In addition, the wom­ en were still married in 1969 and had successfully completed the 1969 interview.18 The second column is the actual distribution of methods used in 1969 for women meeting these same criteria except for the one year of age overlap. We estimate that 42 percent of these women would have been using no method or the rhythm method by 1969, and that 38 percent of women using any method would have been using rhythm. These "ex17 We also made the comparisons for women married at least 3 years but it made little difference. 18 These two criteria could not be applied to women 40-44 in 1965 who were not included in the 1969 reinterview.

CHAPTER Vlll

pected" values are based on the assumption that the only factors in­ volved are the additional several years of marriage and all that this implies for fertility. It does not assume any secular trend in conformity to Church norms on birth control. In 1969, we observe even lower pro­ portions conforming: 36 percent of all women reported using no method or the rhythm method most recently and 34 percent of women who had used any method reported rhythm as the method used most recently. These differences indicate that the trend toward nonconformity had not only not been reversed but had not even been arrested. Table VIII-13 also shows the dominant role of the pill in methods used between 1965 and 1969. There was a tremendous increase in the reliance of Catholic women (married more than 4 years) on this method, from 14 percent in 1965 to 37 percent by 1969. In terms of net shifts, most of this increase (17 of the 23 percent) resulted from shifts from other methods, while the other 6 percent came from those who had pre­ viously used rhythm or no method. For the first time we can record the fact that the pill is not only more popular among Catholics than all other nonapproved methods combined but that, even allowing for the slight sample bias in favor of the pill (Table VIII-10), it may now exceed rhythm as the most popular Catholic method. Recent Changes Among Individual Couples

Our analysis thus far has focused on aggregate comparisons, that is, we have compared samples of women at different times in terms of the distribution of methods used. We turn now to an examination of patterns of change among individual couples. The data for this analysis are drawn exclusively from the 1969 Study. We shall first examine current practice and the overall extent of recent changes in methods used. There are 459 still-married Catholic women (classified as fecund in 1965) who answered our questions. Their current contraceptive prac­ tices are described in Table VIII-14. The majority of women—70 per­ cent—are using some form of contraception; among these, only onethird are using rhythm.19 Which methods did these women use pre­ viously? Which methods were used by women who are now pregnant, or trying to get pregnant, or not using contraception for other reasons? These questions are answered in Table VIII-15. There we show the methods previously used both by women not currently using any method 19 If we estimate the use of rhythm by classifying women not currently using any method by the method they intend to use in the future and add them to the distribution of women using contraception, a slightly higher proportion (37 percent compared to 34 percent) would be using rhythm.

Table VIII-14.

Current contraceptive 'practice for Catholic women in 1969.

Percent Hot Using Contraception

30.1

Pregnant Trying to become pregnant Subfecund Other reasons Using Contraception

7.0 3.9 4.8 14.4 69.9

Rhythm Pill Other Methods

23.7 27.9 18.3

Percent Total

100.0

Number of Women

459

Table VIII-15. Current and most recent contraceptive practice for Catholic fecund women in 1969.

Current Practice

Most Recent Practice

No Method

Percent 27.0

No Method Rhythm Pill Other Methods Method

10.7 6.0 7.7 2.6 73.0

Rhythm Rhythm Only Pill Other. Methods

20.9 2.1 1..2

Pill Only Rhythm Other Methods

11.2 12.3 6.3

Pill

Other Miathods Other Methods Only Rhythm Pill

9.1 4.4 5.6

Percent Total

100.0

Number of Women

4303

a

Excludes 7 women whose immediate previous history was not determined.

211

CHAPTER Vlll

and by women currently using rhythm, the pill or other methods. In Table VIII-15 we have excluded the 22 women who are not using any method because of sterilizing surgery or presumed subfecundity. The methods last used by those currently not using any method are the pill, rhythm, and other methods, in that order—the same rank order of use as for women currently practicing contraception. Among women currently practicing rhythm, we observe that the great majority (86 per­ cent of the 104 women) had used only rhythm before; there is very little indication of return to the Church-approved method after the Encycli­ cal. Among those users who had formerly used rhythm, 30 percent had switched to the pill. This tendency for Catholic women who had used rhythm to switch to the pill rather than other methods is underscored by the fact that only 11 percent of former rhythm users are currently using other methods. This preference for the pill can probably be explained not only by the attractiveness of the method per se but also, in the case of Catholics, by the theological ambiguities surrounding it at least in the past several years. Changes Among Individual Couples Since the Encyclical

In the preceding section we looked only at recent changes in contra­ ceptive practice without attempting to specify whether the changes oc­ curred before or after July 29, 1968. It is clear that, even if we had collected our data on a month-to-month basis, some measurement error would be involved. Depending upon women to remember the month they switched to another method, even though the critical time referent is less than a year—and we attempt only to distinguish between before and after the Encyclical—is obviously more perilous. Therefore we have to emphasize that accuracy of recall is a consideration in evaluating the fol­ lowing conclusions. It is of course reassuring that our conclusions are consistent with the pattern of change in Table VIII-15 which does not depend on the recall of when the change occurred. In addition to inquir­ ing about when a couple switched from one method to another we also asked, when appropriate, about their contraceptive intentions—whether they intended to use rhythm, some other method, or no method at all in the future. We have thus assembled all the information we have about the recent past, current practice, and future intentions in order to esti­ mate how many Catholics changed their practice after the Encyclical. Of 238 women who were using the pill or other nonapproved methods at the approximate time of the Encyclical, we identified only 7 women who are now conforming to Church teaching (either by using no method

BEFORE AND AFTER THE ENCYCLICAL

for some reason or by using rhythm) or who intend to conform in the future either by switching to rhythm after they have a child or by not using any method. We might reasonably categorize these 7 women as having been moved in the direction of conformity by the Papal Encycli­ cal, although of course they might have changed their practice for other reasons. There is another group of women who might have been influenced in the opposite way: some may have been expecting a liberalization of the Church position and become alienated by the final pronouncement. We can identify 5 cases of women currently using the pill or some other method who had switched from rhythm since the Encyclical. (This does not include those who began such practice after having used no method before, since we did not collect data on when their contraceptive prac­ tice began but only on when some preceding contraceptive practice stopped.) These 5 women probably represent more of a continuation of the trend, however, than any sudden alienation. And, as we have ob­ served before, we do not know that this "trend" might not have been greater in the absence of the Encyclical, that is, we have no sure way of estimating how many conforming Catholics on the verge of abandoning the rhythm method may have been deterred by the Encyclical. Conclusions

As we have stressed repeatedly, we have little basis for really knowing what proportion of Catholics would have been conforming to Church teaching in 1969 if the Pope had not issued the Encyclical (let alone what might have happened had the Encyclical taken the form of liberal­ izing the Church position). It is possible that the trend toward noncon­ formity would have been even greater than observed and that what we are interpreting as a continuation of the rate of attrition may in fact be a deceleration. This interpretation seems a bit extreme, however. Our most conservative conclusion is that the Papal Encyclical has certainly not effected the slightest reversal in the trend toward nonconformity and has probably not even slowed it down.20 Catholic Women's Attitudes Toward the Encyclical

We were interested not only in what effect the Encyclical had on con­ traceptive behavior, but also in how Catholic women reacted to it. 20 This conclusion agrees essentially with the speculations advanced in John Marshall, "Population Policies in the Light of the Papal Encyclical 'Humanae Vitae'," Proceedings of the International Population Union (London, 1969).

CHAPTER VlU

Toward the end of our telephone interview we asked the following openended question: In July of 1968, the Pope issued his encyclical re-emphasizing the Catholic position against any method of family planning except the rhythm method. How would you describe your reaction to the stand the Pope took against any method of family planning except the rhythm method? The major theme running through the responses to this question is the assertion of the right of individual conscience in the matter of birth con­ trol. One receives the impression that many women had made up their own minds before the Papal announcement and were no longer willing to acquiesce to Papal authority on this question. Repeated use of phrases such as "It's a matter of personal conscience" suggest that the more liberal views expressed during the period of open controversy within the Church prior to the Pope's announcement were a source of encourage­ ment for this independent attitude. Some illustrative responses are: People know how they can take care of their children and educate them. . . . I think it should be up to your own individual conscience. ... I'm a Catholic and practice my religion except for birth control. I don't agree and I happen to be a Catholic. .. . I felt that just because he could debate it, and it was a man-made law—that I'm sorry, I just don't agree. . . . The latest theory in our Church is the conscience method and therefore my conscience is clear. I don't think the Church has any right to dictate how many children you should have. I don't think any church has a right to pry into any­ one's private life. The Church doesn't help raise the children nor pro­ vide for them. It's just not the business of the Church. As far as we were concerned we had no quarrel. . . . It does depend on how it affects individual families. . . . Some families get too large, and they can't afford all the children. . . . It's really up to the individual to decide but my husband and I are satisfied. As some of these responses show, much of the impetus toward such independent thinking seems to have stemmed from the necessity of cop­ ing with the practical realities of life, and such considerations are in fact behind a great deal of the sharpest criticism of the Papal position. Wom­ en whose health is threatened by continued childbearing and those whose families are in a precarious financial condition have frequently answered this question with considerable feeling:

BEFORE AND AFTER THE ENCYCLICAL

I think the Church and the Roman Catholics don't understand. . . . Maybe the priests just don't understand because they don't marry, but you can't just keep on having kids and provide for them. We're a family of six, and my God how many more should we be expected to have? I'll tell you I've had 3 children since I spoke to you. . . . I was always against the use of anything until my doctor informed me that my insides were all messed up. . . . The doctor wants me to take the birth control pill for 3 years because it would be injurious to my health to have any children. . . . I agree with the Pope in anything, but when it comes to health, it's a different story. Well, I didn't really pay any attention to what he said because I had been taking the pill. After all, we are the ones standing at the counter at the grocery store and paying the bills. We should be able to say what we can afford to do and how many children we can afford to have and have the ability to take care of. We should be able to give them the attention they deserve, and you can't do that when you have so many children. Even those whose own immediate circumstances are not necessarily critical may cite the wider social picture as a consideration which can­ not be ignored: I think they were wrong. . . . We're fortunate in being able to afford these children, whereas others have to be able to plan because of eco­ nomic difficulties and [consider] what the conditions of their families are. I was really horrified. . . . Well, I think that with the earning power that the people have now and living in cities rather than in rural areas, your dollar doesn't go as far. People are living in apartments and can't possibly raise these huge families in apartments. On farms they had corn and fruit and vegetables, and it was easier. . . . Another reason is I don't believe that God meant for people to have one child after an­ other when it might impair the mother's health or cost her her life. I just can't believe He's against preventing conception. As one would expect, there is considerable evidence of conflict and confusion in the replies. For some, like the following, the more liberal opinion of a lesser authority presented a way out. I was shocked. . . . I thought he would go the other way and there would be some loopholes. I knew that he couldn't change his stand. I

CHAPTER VIIl

changed my thinking in the last two years. . . . I respect his position and quandary. I did the intelligent thing. . . . I took the advice of my parish priest. [She is currently taking the pills.] At the time it was issued we were using contraceptive methods, and I stopped for 6 months until the Archbishop of Detroit came out about conscience. . . . Somehow his reasons are misunderstood by some peo­ ple. . . . He wants to stress to people that family planning is not the only answer to all the world problems. . . . He does leave room for pri­ vate interpretation. This conflict in authority can be seen in the fact that 1 out of every 4 Catholic women said that at least some of their parish priests disagreed with the Pope on this issue. There is considerable expression of hope for further change. For example: It's the only stand he could take. . . . In principle he's right. In prac­ tice I don't know how it would work out. . . . Well, in practice some kind of help will have to be found for families, either a new develop­ ment of some kind or a new pill that won't be against the teaching of the Catholic religion. I think they are working on one at the present time. . . . Well, I have to say that I agree with him—in principle anyway. It's my religion. I was born into it, and I believe it. I wish they'd change it though; they're changing everything else. But a great many simply express doubt and indecision. Phrases such as "I was disappointed" or "I don't know" appear very frequently and often lead into comments which are highly ambivalent. I was disappointed. . . . I felt that I wanted to use it. . . . I don't want to have any more children. . . . Going against the Church bothers your conscience if you take the pill. It makes it hard if you are a Catholic. I really don't know. . . . It's a thing I can't figure out. . . . I think it's wrong of the Catholic religion to say you shouldn't prevent having children if you can't afford to have them. . . . I really have very mixed feelings over this. . . . I guess everyone has to decide for themself what is right or wrong. I really couldn't tell you. . . . I have no feelings. . . . I have never been asked this question before. . . . I have some thoughts, I don't know

BEFORE AND AFTER THE ENCYCLICAL

how to say them. . . . I can't say whether I agree or disagree. I don't know. Feeling on this subject has evidently frequently spilled over into a gen­ eral reaction toward the Church and its teachings. Women who cannot rationalize the situation in any other way have felt compelled to with­ draw from the Church to a greater or lesser extent. The following are examples of this type of response: I reacted like a lot of other women. . . . I stopped going to church. . . . I couldn't see much sense in it; it seemed like a farce. . . . If all the women who take the pill stopped going to church, there would hardly be any women there, only men and children. . . . I felt I was hypocriti­ cal to go to church and not receive the blessing. . . . I have 13 chil­ dren, and I really can't afford to have any more, so I had to make my choice. . . . I feel badly about it, but I can't help myself. . . . I feel the Pope was wrong in his stand, so I'll take the pill and go to church once in a great while. We've changed religion. I feel it's unjust. . . . He's put people in the position of having children they don't need or want. . . . I don't believe that God feels that we can't prevent children if we can't take care of them. I had made my own decision long before that concerning religion and that aspect. . . . I think it's ridiculous. . . . My whole opinion or feeling about the Church was re-emphasized by that stand. . . . I feel that the Catholic Church does not offer realistic spiritual guidance. . . . I think there are a lot of young mothers experiencing serious emotional prob­ lems because they are torn between wanting to be a good Catholic and wanting to be adequate mothers. . . . I do feel that what is happening in the Church is that couples that are intelligent see that it is more meaningful to provide a happy family life on their own terms and not to become involved with the Church's controversy. . . . Partially be­ cause of this issue, when I married I decided this was unrealistic to me, so I decided to stay away from the Church. Since, I have recently be­ come involved with European Catholics. They seem to handle it more properly; they seem to feel the Church expected them to use their own judgment . . . concerning birth control. Why make a stand divorcing yourself from spiritual learning and depriving my children because of this whole ridiculous stand on birth control? I feel my children would be missing something spiritual if I stayed away from the Church, yet personally I feel they are ridiculous.

CHAPTER Vlll

In recognition of the likelihood of such reactions, many of the more thoughtful respondents express concern over what this whole policy has meant for their Church: We were very unhappy about it. . . . Well, we feel like other methods are acceptable, and we didn't like him not agreeing. . . . I completely disapprove of his stand. . . . He's hurting the whole Church. . . . So many married couples have been using other methods, they have to by the law of reason; they are going against their church and therefore breaking away. I don't know.. .. We were disappointed. . . . Because we were hoping we would not have children. . . . Some people, no matter what he says, their minds are made u p . . . . They are going to go against h i m . . . . W e were hoping that he would condone some other method to stop or limit it. . . . My younger brother is a priest and he believes that the young should limit their families. . . . He's younger and newer, and this is what he believes. . . . People used to go oftener to receive com­ munion, but I believe because of the Pope's stand it is keeping young people from the sacraments. I don't know. I don't have much of a reaction. I don't think it was very well interpreted as the Pope intended it to be. . .. I do agree. It would have been more of a conflict if he had reversed his opinion, and he would have compromised the whole religion. It would have been con­ fusing. I don't feel it was his decision solely. It's such a part of the religion that it would compromise the whole religion had he made another decision.... He has to consider others. Well, I feel it's very unrealistic, and I believe it's more a question of Papal authority rather than the method of birth control. . . . I believe Catholicism will be changed eventually. . . . I just don't think the question is being weighed on its merit. It's being weighed within the Church right now as a matter of observing the Church's rules, and as Catholicism is liberalized, there will be a more realistic approach to the population problem. Finally there are the responses of those who accept the Church doc­ trine unquestioningly, statements made emphatically and without apology. These last appear to be the only ones to whom it presented it­ self as a simple matter. Well, as a Catholic I go along with him. ... I was properly brought up that way.

BEFORE AND AFTER THE ENCYCLICAL

I agree with him. ... It's the only method that is right. . . . I feel very strongly about this.... I feel it is right. I was expecting it. I didn't think it would be any different. . . . Because it's not just his law; it's God's law. Prevention of children is up to God. There is no excuse for taking the pill. If you don't want children, then you just have to be careful. . . . Use the rhythm method or abstain. I just agree with it. I could never see the Pope saying anything else. I think that's very right. . . . I think that's the way the Lord meant the earth to be. How do these reactions balance out? The attitudes were coded into three basic categories: for, where the woman's response indicated full approval or acceptance of the Pope's statement; mixed, where the re­ sponse was not a clear-cut positive or negative, where she makes both favorable and critical comments; and against, where the comments indi­ cate disagreement without any reservation. The distribution of responses according to this code are shown in Table VIII-16. It is obvious that the preponderance of Catholic opinion is clearly against the Papal decision. Table VIII-16.

General Catholic attitude toward the Papal Encyclical.

Attitude

For

Percent

16.3

Mixed

18.7

Against

60.2

Sot Classifiable

4.8

Percent Total

100.0

Niimber of Women

455

Are Catholics of different ages, educational backgrounds and degrees of involvement with their Church uniformly against or in favor of the En­ cyclical or are there differences in reactions which characterize various segments of the Catholic population? This analysis appears in Table VIII-17. In general the older a woman is the more likely she is to favor the Encyclical, but even among the oldest women in the sample (39-43) only 24 percent expressed themselves unambiguously in favor of the

CHAPTER VlII Table VIII-17.

Catholic attitudes toward the Papal Encyclical, by age,, education, and frequency of attending Mass.

For

Percent Mixed

Total

17

20

63

100

433'

Age (1969) Under 24 24-28 29-33 34-38 39-43

4 13 17 18 24

16 13 26 20 22

80 74 57 62 54

100 100 100 100 100

25 110 83 120 95

Education (1965) College 4 College 1-3 High School 4 Grade School

18 8 19 17

35 31 15 19

47 61 66 64

100 100 100 100

34 51 248 100

Mass Attendance (1969) Several times a week Once a week Several times a month Once a month Several times a year Once a year or less

42 20 7 4 3 6

19 22 23 17 17 3

39 58 70 78 79 92

100 100 100 100 100 100

26 287 30 23 29 36

Against

Percent Total

Number of Women

Pope's position. At the opposite extreme, among the youngest wom­ en (under 24) 80 percent recorded unambiguous opposition to the Encyclical. The amount of education a woman has received bears some relation­ ship to her attitude toward the Encyclical, but it is not a straightforward simple one. Women with 4 years of college education appear to be least opposed to the Encyclical perhaps because of the concentration of more religious Catholics in that educational class. Women who have attended college are much more likely to register a mixed response than those with less education. The association of attitude with the religiousness of Catholic women (represented by the frequency with which they attend Mass) is much stronger and clearer than the relationship with education. The least opposed are those who attend Mass more than once a week (39 percent against); the women who attend once a year or less are almost all (92 percent) opposed to the Encyclical. The bulk (two-thirds) of Catholic women in the sample attend Mass once a week; among these regular

BEFORE AND AFTER THE ENCYCLICAL

churchgoers only 20 percent are clearly in accord with the Pope's position. Attitude and Behavior

The final question we pursue in our study of the effects of the En­ cyclical is the consistency of the woman's attitude toward it with her contraceptive behavior. In Table VIII-18 we present the proportions of Table VIII-18.

Attitude

Total

Catholic attitude toward the Encyclical and the proportion of Catholic couples using nonapproved methods3 o£ contraception.

Percent of Total Using Nonapproved Methods-

Percent of Users Using Nonapproved Methods

Number Total Users

52.3

64.7

429

8.4

12.0

71

50

Mixed

49.0

58.0

96

81

Against

65.3

79.5

262

215

For

a

346'

Women currently pregnant or trying to get pregnant assigned to method used before. "Users" include women who were using a method before becoming pregnant or trying to become pregnant.

women using nonapproved methods of birth control by their attitude toward the Encyclical. Only a small fraction (8 percent) of women in favor of the Encyclical are not conforming to their Church's teaching; among those with mixed feelings, about half are conforming; and, among women who registered disapproval of the Encyclical, the majority are using nonapproved methods. Nonetheless a substantial minority of those opposed to the Encyclical were using either the rhythm method or no method (excluding women currently pregnant or trying to become preg­ nant). In general, however, there seems to be considerable integration of attitudes and behavior.

CHAPTER IX

Fertility Planning AN IMPORTANT aim of any fertility inquiry is assessment of the extent to which, couples report success or failure in their attempts to achieve their reproductive goals. Although the general topic encompasses failure (because of infecundity) to become pregnant when that is desired, our attention here is confined to failure to prevent pregnancies from occur­ ring if they are unwanted. The objective is to present a procedure for measuring the proportion of couples who have failed in contraceptive efforts to control the number or the timing of their pregnancies, and to discuss the results of applying the procedure to responses from a sample of wives, cross-classified by race, religion, and education. The questions of primary importance for this chapter are those asked of every respondent concerning the intervals preceding each of her preg­ nancies, if any, and the open interval (the one terminated by the inter­ view). First it was determined whether the couple had employed any method of birth control during the interval. For those reporting no method, we asked: "Was the only reason you did not use any method then because you wanted to have a baby as soon as possible?" If the re­ sponse was affirmative, the pregnancy was classified as a nonfailure, but not as a success (from a planning standpoint). If the response was nega­ tive, the pregnancy was classified as a failure. For those reporting use of some method of contraception during the interval, the key question was: "Under which of these circumstances did this pregnancy occur?" The respondent answered by selecting from a card one of the following three options: 1. While using a method and did not want to become pregnant at that time. 2. While not using a method but did not want to become pregnant at that time. 3. When stopped using a method in order to have a child. Pregnancies for which either of the first two options was recorded were classified as failures. Pregnancies for which the third option was recorded were classified as timing successes. For each pregnancy classified as a failure, whether through nonuse or use, the respondent was also asked the following two questions: (1) "Before you became pregnant this time, did you want to have a(nother)

CHAPTER IX

child sometime?"; (2) "Did your husband want to have a(nother) child sometime?" If the answers to both questions were "Yes," the pregnancy was classified as a timing failure. A negative answer to either of these questions resulted in the classification of the pregnancy as unwanted. In more than 80 percent of the (most recent) pregnancies classified as unwanted, neither husband nor wife wanted a(nother) child some­ time. We decided (following the precedent of Whelpton et al.)1 to con­ sider as unwanted that small minority of pregnancies concerning which the husband and wife apparently disagreed. We investigated variations by race, religion, and education in the extent of disagreement between spouses on whether the last pregnancy was wanted, and all variations were small. Since the wife was the informant for her husband as well as herself, we do not consider our data well suited to further exploration of this topic. The above procedure provides a classification of all pregnancies, as follows: (a) unwanted; (b) timing failure; (c) timing success; (d) nonfailure (no contraceptive used because a pregnancy was wanted as soon as possible). Our focus is the overall performance of the respondent up to the interview date. We excluded 31 respondents (less than 1 percent of the total) whom we classified as nonplanners: these couples had never used contraception, wanted all of their pregnancies as soon as possible, and (although apparently fecund) stated that they would never use con­ traception. They were concentrated in the lower parities. If they re­ mained fecund, they might subsequently be compelled by recurrent pregnancies to change their minds about the use of contraception. But for the present they are nonplanners because they cannot be classified as either successful or unsuccessful so long as they eschew altogether the idea of planning. FERTILITY PLANNING STATUS Statuses for Those Who Intend No More Children

In determining the planning status of the respondent on the basis of the classification of her several pregnancies, we first divide the sample (exclusive of nonplanners) into the two-thirds (67.6 percent) who in­ tend no more children and the one-third who say they intend to have more children. The reasoning behind this division was that couples who intend more children are, by the nature of the case, not yet exposed to the risk of excess fertility. 1 P. K. Whelpton, A. A. Campbell, and J. E. Patterson, Fertility and Family Planning in the United States (Princeton: Princeton University Press, 1966).

224

FERTILITY PLANNING

The classification of couples who intend no more children is displayed in the stub of Table IX-1. Respondents are dichotomized first into those who had never and those who had ever been pregnant. The latter are further dichotomized into the categories "number failure" and "no num­ ber failure," based on the classification of their most recent pregnancy Table IX-1.

Fertility Planning Status (FPS) distribution of wives who plan no (more) children, proportion currently sterile, and proportion of the nonsterile not currently using contraception, by FPS

Proportion FPS Distribution (%)

FPS Total3 O Pregnancies 1+ Pregnancies

Proportion

of Nonsterile

Sterile (7°)

Not Using (7„)

100.0

27

16

3.3

71 26

38 16

96.7

Humber Failure

31.2

26

14

No Number Failure

65.5

26

17

34.4 31.1

23 30

16 19

1+ Timing Failures O Timing Failures 1+ Timing Successes O Timing Successes

20.7

23

6

10.4

44

54

Unweighted N = 3239.

as unwanted or wanted. The latter group are still further dichotomized into those who had ever had a timing failure and those who had never. Finally, those without a failure are dichotomized by whether or not they had ever timed a pregnancy successfully. Successful timing required that at least one pregnancy occurred after contraception was stopped in order to conceive; all others without failure said that every pregnancy occurred without use of contraception in the preceding interval because all were wanted "as soon as possible." In order to assess the prospect of subsequent change in the distribu­ tion of these women by planning status, we first calculated the proportion who were definitely sterile at the time of the interview. There are two noteworthy deviations from the total proportion of 27 percent: (1) the proportion currently sterile is (not unexpectedly) very high among the

CHAPTER IX

never-pregnant; (2) a substantially larger than average proportion of those recorded as having neither successes nor failures (wanted all preg­ nancies as soon as possible) is sterile. This may be a consequence of their tendency not to use contraception in the open interval. In this way, unlike users, they put their fecundity to the test. On the other hand, it may signify resort to sterilization as a method of fertility control. For those who are not definitely sterile we calculated the proportion who are apparently exposed to the risk of an unwanted pregnancy (and thus to the risk of reclassification as a number failure) because they are not currently using contraception. Deviations from this proportion for the total (16 percent) are noteworthy in three categories: (1) it is sub­ stantially higher among the never-pregnant, perhaps because they are in fact sterile but not so identified by our (rather conservative) criteria; (2) it is quite low for those classified as having at least one timing suc­ cess (and no failure) in their previous record; (3) it is very high for those with neither successes nor failures in their previous record. It seems to us that this last observation represents evidence for the in­ validity of some responses to the question (asked of those who reported nonuse in any particular interval) of whether or not they wanted that pregnancy as soon as possible. Note that, had these women stated their intent never to use contraception in the future, they would have been classified as nonplanners and excluded from the analysis. These women seem to be prime candidates for reclassification as number failures in due course. Pursuing this question further, we subdivided the couples classi­ fied as number or timing failures in Table IX-I on the basis of whether or not they had used contraception before the end of their last preg­ nancy. Of those who had, the proportion (of the nonsterile) not currently using contraception was 8 percent; of those who had not, the proportion was 63 percent. (These correspond with the proportions in the final two categories in column 3 of Table IX-1.) The conclusion is that whether or not a couple is risking a future failure depends on whether they have used contraception before the end of the last pregnancy, inde­ pendently of whether they called their prior nonuse intentional or un­ intentional. We conclude that the nonsuccess category represents either a concealment of failure or, at the least, a behavior pattern which carries high risk of eventual failure. Data such as these represent a strong argu­ ment for reorienting inquiries like ours from the retrospective design to a longitudinal format, in which statements of intent are collected before rather than after the events in question.

FERTILITY PLANNING

In devising indices of the proportion of couples who have failed in their efforts to plan fertility, we have restricted the calculation to those exposed to the risk of failure of the type specified. The results are pre­ sented, for couples who intend no more children, in Table IX-2. The first panel contains the proportions of ever-pregnant women classified as number failures because their last pregnancies were unwanted (by either or both of the spouses). The decision to exclude never-pregnant women from the denominator was based on the consideration that, although some of them were probably childless by choice (and employed contra­ ception effectively to that end) the majority were in fact sterile and thus not exposed to the risk of an unwanted pregnancy. From another stand­ point, we are basing the planning status classification on data for closed intervals; for never-pregnant women we have only the open interval. Approximately one-third of these women are number failures. The discrepancy between Negro and white couples is large—most Negro cou­ ples fall into this category—but there is only a negligible difference by religion (within the white subsample). The difference by religion may not remain negligible: the proportion currently risking a number failure by not using, even though nonsterile, is 12 percent for white non-Catho­ lics but 22 percent for white Catholics. There is an inverse relationship between the probability of number failure and years of schooling (of wife), a relationship which also holds within the Negro and the white non-Catholic subdivisions. The relationship of the probability of a num­ ber failure with education is not as pronounced as that with race; for example, it is distinctly higher for Negro wives with at least some college education than it is for white wives who failed to complete high school. For white Catholics, however, the relationship of the probability of a number failure with education, although weak, is in fact direct. The contrast between the role of education among Catholics and among others has been observed in various analyses of the National Fertility Study. For example, the relationship between education and four meas­ ures of fertility was inverse for white non-Catholics, but tended to be U-shaped for white Catholics. Pertinent to the present question (of the higher incidence of number failure among college-educated Catholics) is the fact that this group, which is drawn from a more religious back­ ground than Catholics of less education, has the highest proportion using rhythm only (38 percent as against 31 percent and 10 percent) and the lowest proportion using other methods (44 percent as against 52 percent and 58 percent) among the three education groups (see Chapter VIII).

Proportion with last pregnancy unwanted.

174

56

651

421

63

55 51 75

56

84

57 56 60

58

285 237 48 27

236

314

1459' U12 347

1719

74

721 550 171

799

135

453 325 128

606

87

70 71 66

72

Unweighted Denominators

84

60 59 64

62

Percent

Conditional Probability of Timing Failure Education Total >H.S.4 H.S.4 CH.S.4

8

28 29 26

26

22

34 37 17

33

8

31 32 28

30

6

20 18 24

17

Probability of Successful Planning Education Total >H.S.4 H.S.4' CH.S.4

Product of the complements of the proportions in the first two panels.

Includes other nonwhites.

c

d

^ Of those who wanted their last pregnancy, and did not want all pregnancies as soon as possible, the proportion with at least one timing failure.

a

Negro

899 660 239

1125 833 292

412 333 79

2436 1826 610

White Non-Catholic Catholic

1354

1309

3133

Total

470

41

52

Negro 55

34 37 28

28 27 30

25 24 32

30 30 29

White Non-Catholic Catholic

48

38

29

26

32

Total^

Race and Religion

Probability of Number Failure Education Total >H.S.4 H.S.4 CH.S.4

SL b Table IX-2. Probability of number failure; conditional probability of timing failure; probability of successful planning0 for ever-pregnant wives (exclusive of nonplanners) who intend no more children, by race, religion, and education.

FERTILITY PLANNING

Since rhythm is less likely than other contraceptive methods to prevent an unwanted pregnancy, it is not surprising that higher failure propor­ tions are observed within the college-educated Catholic group. In Table IX-2, the link between the probability of number failure and planning efficacy (percent reduction in the probability of conception per exposed ovulation) is not a direct one, because the former measure ex­ cludes the time dimension from its calculation. Our attempts to measure planning efficacy inclusive of the time dimension are described in Chap­ ter XI. For the present, a suggestion can be made of the direction of difference between the measure of failure and the measure of efficacy. The interval between the birth of the last intended child and the inter­ view (and thus the interval of exposure to the risk of a number failure) was approximately 6 years for white Catholics but 8 years for white nonCatholics, essentially because the former planned for 20 percent more children, and that takes more time. Accordingly, if the proportion fail­ ing is put on a per annum basis, it is approximately 5 percent for Catho­ lics as against approximately 4 percent for non-Catholics. With respect to the discrepancy between the proportions of Negroes and whites with number failures, the implicit difference in contraceptive efficacy is even greater (11 percent per annum as against 5 percent), not because the Negro wives desired more children, but because they were more than a year younger on the average than the whites. These remarks are not in­ tended to denigrate the observations reported here: our intention is to report the prevalence of couples with failures, and these failures result from a combination of efficacy in planning and exposure to risk. In the second panel of Table IX-2 we show, for women who have not had a number failure, the probability that their record includes a timing failure. Excluded from the denominator of this calculation are women who did not use contraception in the interval prior to their last preg­ nancy because they claimed to have wanted all of their pregnancies as soon as possible. (Such women constitute some 16 percent of those with­ out a number failure. Their prevalence by race, religion, and education shows the same general pattern as the prevalence of failures. This is fur­ ther reinforcement of the proposition that their reporting is suspect.) The conditional probability of timing failure is 62 percent for the total sample. The difference between Negroes and whites is large, that be­ tween white Catholics and white non-Catholics is small, and the relation­ ship with education (again excepting college-educated Catholics) is negative. Granted that these differentials are all in the expected direc­ tions, the remarkable circumstance is that the differentials are so small

CHAPTER IX

relative to the average. The high prevalence of timing failure, like that of number failure, is a characteristic of all women in the United States, irrespective of race, religion, or education. As a way of summarizing, we have prepared the data shown in the third panel of Table IX-2. They represent the product of the comple­ ments of the proportions in the first two panels. Thus they express the combined effect of number and timing failures, in a positive rather than a negative way, as the probability of successful planning. Only 26 per­ cent of the wives fall into this category: the range of variation is from 37 percent for the white non-Catholic women with at least some college education to 6 percent for the Negro women who did not complete high school. The pattern of differentials corresponds to that described for the first two panels of the table. However, this measure is somewhat less use­ ful than its component parts because it is unreasonable to weight equally number and timing failures; the former have much more serious conse­ quences than the latter. Furthermore, a timing failure which occurs de­ spite the use of contraception is not necessarily a complete failure, be­ cause the couple may have achieved a substantial part of their desired delay, even though not all of it, whereas a number failure is by definition a complete failure. Statuses for Those Who Intend More Children

For the one-third of our sample who intend to have more children, we have prepared Tables IX-3 and IX-4, along lines parallel to those for Tables IX-I and IX-2. Since these women consider their childbearing to be incomplete, none of them is classified as a number failure. Other­ wise the respondents in this subdivision are dichotomized exactly as they were in Table IX-I; the definitions are the same. In order to assess the likelihood of a change in planning status with the occurrence of the next pregnancy, we have first calculated the pro­ portion of these women not exposed to the risk of a timing failure, that is, those who report no use in the open interval because they claim to want the next pregnancy as soon as possible. Of these women, the only ones who could experience a change of status would be those who have never been pregnant. Of the remainder (those at risk) we have calcu­ lated the proportion not using contraception currently. (The numerator does not include women who had been using in the open interval, but had then stopped in order to conceive and had not yet done so by the time of the interview.) Twenty percent of this sample are apparently run­ ning the risk of a timing failure with their next pregnancy. The only sub-

FERTILITY PLANNING Table IX-3.

Fertility Planning Status (FPS) distribution of wives who intend at least one (more) child, proportion not exposed to the risk of timing failure, and proportion of those exposed to the risk not using contraception, by FPS.

FPS Distribution a)

FPS Totala 0 Pregnancies 1+ Pregnancies 1+ Timing Failures 0 Timing Failures 1+' Timing Successes 0 Timing Successes

a

Proportion Not Exposed

(%)

Proportion of Exposed Hot Using (7·)

100.0

11

20

18.1 81.9

27 8

21 20

43.9 38.0

5 10

17 25

22.8 15.2

7 15

13 45

Unweighted N = 1540.

group which departs substantially from the figure for the total is women with no timing successes. Once again the implication is that their reports of planning status for their previous pregnancy or pregnancies are sus­ pect. (The comparative prevalence of this subcategory by race, religion, and education shows the same suggestive pattern mentioned in the dis­ cussion of Table IX-2.) As before, we classified the timing failures ac­ cording to whether or not they had used contraception prior to the ter­ mination of their last pregnancy, and found that 9 percent of the users, but 47 percent of the nonusers, were risking further failure (in compari­ son with 13 percent and 45 percent for the same measure among the non-failures). For those who intend to have more children, two-thirds report at least one timing failure already. (The never-pregnant are excluded from the denominator because the only information we have for them is their open interval. See Table IX-4.) The pattern of differentials is the same as be­ fore: a large gap between Negroes and whites, a smaller difference be­ tween white Catholics and white non-Catholics, and an inverse relation­ ship with education (within subgroups by race and religion as well as for the total subsample). Once again, the most remarkable finding is the generally high level of failure, compared with which the differentials by race, religion, and education are secondary. As with the previous tables there would be problems in interpreting these data as direct measures of planning efficacy. The probability of a woman having a pregnancy in the

CHAPTER IX

"timing failure" category obviously depends on the number of preg­ nancies the woman has experienced. (There is also dependence on the length of exposure; none of the measures in the tables in this chapter takes the time dimension into account.) For the respondents whose be­ havior is reported in Table IX-4, the mean number of pregnancies is 2.1 for white Catholics and 1.4 for white non-Catholics. The failure probability per pregnancy is in fact higher for white non-Catholics (0.49) than it is for white CathoUcs (0.44). Since Negroes and whites have the same number of pregnancies (1.6), there can be no reversal in that comparison: the failure probability per pregnancy is 0.66 for the Negroes and 0.47 for the whites. Table IX-4. Probability of timing failure for those exposed to the riska who have been pregnant and intend more children, by race, religion and education.

Race and Religion Total

> H.S. 4

Education H.S. 4

< H.S. 4

Probability OO of Timing Failure Total

67

60

65

76

White Non-Catholic Catholic

64 61 70

58 52 75

62 58 67

73 73 74

Negro

82

73

74

93

Unweighted Denominators Total

1028

221

507

300

White Non-Catholic Catholic

814 512 302

178 127 51

426 249 177

210 136 74

Negro

198

37

78

83

a

Excludes those who wanted all pregnancies as soon as possible. Includes other nonwhites.

Some Methodological Reservations

Although the failure levels reported above are much higher than we expected, it is our judgment that they underestimate the true level of fail­ ure in the population sampled. The first consideration is the reliability of reporting. It seems plausible that many women would report a preg-

FERTILITY PLANNING

nancy as wanted, after the fact, even though the pregnancy was not wanted—either then or ever—before the fact. Admissions of failure are likely to have been reduced by embarrassment at confessing inadequacy and by rationalization, as well as by a genuine change in judgment based on experience with the previously unwanted child. The passage of time may lead women to claim as successes pregnancies which might at an earlier stage in life have been admitted to be failures. Secondly, it is generally accepted that many pregnancies end in illegal abortions, yet our interviews uncovered very few. Presumably our sam­ ple includes women who would have had pregnancies classified as fail­ ures (from a planning standpoint) but who avoided this classification by aborting (as a second line of defense against the unwanted pregnancy) and not reporting that pregnancy at all. Certainly there are quite a few mysteriously long interpregnancy intervals in our records. On the other hand, some of the failures reported did not in fact result in an unwanted birth, because the pregnancy culminated in a fetal death rather than in a live birth. For completeness' sake, however, it should also be noted that some pregnancies should really be classified as multiple failures, since they ended in more than one live birth. Thirdly, our classification of those with a number failure was based on the most recent pregnancy being unwanted. A significant proportion of these women have in fact experienced more than one unwanted preg­ nancy. Since we asked each woman the questions determining whether or not the pregnancy was ever wanted for every pregnancy classified as a failure, we were able to observe, not infrequently, that women reported unwanted pregnancies (in this sense) followed by wanted pregnancies. Perhaps this is partly a consequence of a change in the conditions in which the wife finds herself (including, in some cases, a change of hus­ band), but perhaps it is a misunderstanding of the question. (The re­ sponse may signify that a particular child was unwanted, on the basis of experience with that child.) Finally, it is evident from Tables IX-I and IX-3 that most of our women are still eligible for further fertility, and therefore for further failures, and that a significant proportion of them are courting such fail­ ures by not using contraception despite their intentions. Moreover, use of contraceptives is insufficient to guarantee achievement of the reproduc­ tive goal. Among those who do not intend to have more children, 86 per­ cent of the number failures and the same proportion of the timing fail­ ures reported that they had used contraception in at least one interval pre­ ceding pregnancy. Among those who intend to have more children, the

CHAPTER IX

comparable proportion is 79 percent. The subject of the efficacy of con­ traceptive use is treated in detail in Chapter XI. There is one small reservation, however, concerning the argument that the probabilities of failure reported here are underestimates of the "true" situation. As is explained more completely in Chapter XI, there is, in any sample which is cross-sectional by age, a bias toward the overrepresentation of women who arrive at any parity at a younger (as distinct from an older) age. Accordingly, a measure, like fertility planning status, which is based on experience over all pregnancies up to the date of interview, will reflect more the experience of those having pregnancies earlier than those having pregnancies later. Since there is a positive correlation be­ tween earlier pregnancies and planning failures, the effect of the crosssectional bias is to produce a higher estimate of failure, ceteris paribus, than would be the case for the average cohort whose entire reproductive age span was examined. Summary

Respondents have been classified by fertility planning status on the basis of the identification of their pregnancies as unwanted, timing fail­ ure, timing success, or nonfailure. Of those who do not intend to have (more) children, there are five possible classes: (1) never-pregnant; (2) number failure; (3) at least one timing failure (but no number fail­ ure); (4) at least one timing success (but no failure); (5) neither suc­ cess nor failure. Among those who have been pregnant, the probability of a number failure was 32 percent, and the conditional probability of a timing failure for the rest was 62 percent. In combination these yield a probability of completely successful planning of 26 percent, an extraor­ dinarily low level for a population generally considered to be relatively sophisticated in contraceptive matters. For those who intend to have more children, a four-fold classification has been established: (1) never-pregnant; (2) at least one timing fail­ ure; (3) at least one timing success (but no failure); (4) neither success nor failure. The probability of a timing failure, among those exposed to the risk, was 67 percent. Given the fact that the average respondent who intends to have more children has experienced fewer than two preg­ nancies, the probability of planning failure is high. Those respondents who reported that they had used no contraception prior to any of their pregnancies but claimed that the only reason for nonuse was that they wanted the pregnancies as soon as possible pose a difficult classification problem. Because a high proportion of these

FERTILITY PLANNING

women are currently running the risk of a failure in number or in timing (as high as the proportion of women who did not previously use contra­ ception but admitted failure) and because their representation by race, religion, and education resembles that of the failures, it seems plausible to infer that this particular response has low validity, and that at least many of them deserve reclassification as failures. Furthermore, the like­ lihood of some misstatement on questions as sensitive as those concern­ ing competence in planning and attitudes toward one's children, and the fact that most of the sample will continue to be exposed to the risk of failure for some years, among other reasons, prompt the conclusion that the failure probabilities reported here are underestimates, perhaps by a considerable margin, of the true prevalence of failure for completed families. For all measures of planning failure there are distinct differentials by race, religion, and education: higher failure rates for Negroes than for whites, for white Catholics than for white non-Catholics, and for wives with less education (except among Catholics). Nevertheless, these varia­ tions seem secondary in importance to the finding that there is a substan­ tial lack of success in fertility planning across the entire sample, regard­ less of race, religion, or education. An increase in regulatory com­ petence, whether by developing better methods, increasing knowledge of available contraceptive procedures, or practicing greater diligence in their employment, would have considerable consequences for the quan­ tity and time pattern of fertility in the United States. THE TREND IN FERTILITY PLANNING: 1955, 1960, 1965 The principal question to be asked in this section is whether or not American couples have in recent years become more successful in con­ trolling their fertility. We are interested in the trend in the incidence of excess fertility, and in the changing differences in fertility planning suc­ cess by religion, race, and socioeconomic class. Attempts in the 1960 and 1965 Studies to improve upon the 1955 basis of the measurement of fertility planning success have impaired comparability across the three studies. Some of these arc conceptual changes and some result from differences in questioning procedures. Al­ though the authors of the 1960 Study reproduced the classification sys­ tem used in 1955 and we in turn have attempted to produce a system for the 1965 data which would be comparable to that of 1960, some un­ avoidable differences remain. We confine our comparisons here to the two planning categories which seem to us most comparable through

CHAPTER IX

time: couples with histories categorized as "completely planned" (the occurrence of every pregnancy was the consequence of deliberate inter­ ruption of contraception in order to conceive) and those categorized as having "excess fertility" (at least the last pregnancy was not wanted at the time of conception). The proportion of white couples with completely planned fertility changed very little between 1955 and 1965 (Table IX-5); an increase between 1955 and 1960 in the percent completely planned, parity by parity, was largely counterbalanced by an upward shift in the parity dis­ tribution; parity-specific changes between 1960 and 1965 were negligi­ ble. Similarly, the proportions with excess fertility, following a rise be­ tween 1955 and 1960, seem to have changed little during the five years between 1960 and 1965. Both of these generalizations are supported by the detailed intercohort trends presented in Table IX-6. Some of the complexities of these measurements should be described because they seriously affect the interpretation of trends. A change in the proportions of completely planned fertility or excess fertility would seem on the surface to reflect the extent of contraceptive practice, the level of motivation to control fertility, the effectiveness of the contraceptive methods used, and the like. There are, however, two other less obvious factors involved. The first is simply methodological: generally speaking, those reported as having excess fertility at, say, parity four, are couples who intended to stop with their third child. The extent of planning fail­ ures among couples who intend to have three children should be meas­ ured by computing the proportion of all those wanting three children who failed to prevent a fourth birth rather than, as in Table IX-6, com­ puting the proportion in parity four who characterize themselves as hav­ ing excess fertility. The second factor relates to the number of children planned and the times at which they are born. A couple who intends to have a larger family has a shorter exposure to the risk of an unplanned or unwanted pregnancy than a couple who wants a smaller family, ceteris paribus. Furthermore, even among couples desiring the same number of chil­ dren, those who marry earlier and have shorter intervals between succes­ sive children lengthen the period of exposure subsequent to the birth of the last wanted child. Although intended parity increased between 1955 and 1960, it does not appear to have changed between 1960 and 1965. There is, however, indirect evidence that the time pattern of childbearing shifted toward the younger ages, and that we could consequently expect a rise in excess fertility even without any change in the extent of control

24

18

9

3

1

2

3

4

5+

1

5

12

21 8

6

6 51

27

19

25 11

11

46

41

28

11

17

49

35

23

10

18

Percent Excess Fertility 1955 1960 1965

20

Percent Completely Planned 1955 1960 1965

19

otal

Parity

206

263

190 190

682 499

843 468

2414

2713

321

371

614

759

2919

Ntimber of Couples 1955 1960 1965

Table IX-5, Percent of white couples 18-39 with completely planned fertility and percent with excess fertility, by parity: 1955, 1960, 1965.

C H A P T E R IX Table IX-6.

Birth Cohort

Percent of white couples with completely planned and with excess fertility, by wife's age and birth cohort: 1955, I960, and 1965.a

Age 20-24

25-29

of

Wife 30-34

35-39

40-44

Percent Completely Planned 1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

22 25

21

26 26

18

17 21 30

13

16

18

17

17

Percent Excess Fertility 1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

17 11

16 16

15 23 23

20

25 25

25

Number of Couples 1916-20 1921-25 1926-30 1931-35 1936-40 1941-45

a

714 464 440 641

600

748 624 762

695 677 741

572

821

651

Estimates for 1955, 1960, and 1965 appear on the upper, middle, and lower diagonals respectively.

over unwanted childbearing. Accordingly, we interpret the constancy of the proportion of couples with excess fertility to mean an improvement in fertility control within this population, an improvement just sufficient to compensate for the greater risks of excess fertility associated with an earlier time pattern of childbearing. In addition to these questions of interpretation, we should mention the problems of the reliability of measurement. For example, the increase with age in the percent of the 1916-20 cohort with completely planned fertility (Table IX-6) suggests that with the passage of time older wom­ en may have a tendency to reclassify unplanned pregnancies as planned. In order to assess the reliability of such data, we independently coded the responses to the reinterview and examined the consistency of the clas­ sification (see Table IX-7). The first interview produced an estimate of 16 percent in the completely planned category while the second inter-

FERTILITY PLANNING Table IX-7. Consistency of the fertility-planning classification used in 1965.

First Interview Completely planned

Completely Planned 9.3

Partly Planned

Excess Fertility

6.5

Percent of Couples 16.4

60

7.1

62.1

238

15.5

21.5

84

0.6

Partly planned

9.7

45.3

Excess fertility

0.9

5.1

Percent of couples

19.9

56.9

23.2

100.0

Number of couples

71

222

Number of Couples

89

view with the same women yielded an estimate of 20 percent; only 9 per­ cent were classified as completely planned in both interviews. The clas­ sification of excess fertility is somewhat better, but a considerable amount of inconsistency remains. In view of this unreliability one should proceed cautiously with the interpretation of trends based on small differences.2 Trends Among Different Social Groups

Although the evidence suggests little recent change in fertility planning among American couples in general, it is possible that there may be compensating internal changes among different strata of the population. It is also interesting to consider whether Americans are becoming more or less alike in their fertility-planning behavior. Looking first at religion (Table IX-8) we see little evidence of any difference in trend for Protestants and Catholics, except that there may have been a small increase in the proportion of Catholic women with completely planned fertility. The data for the various socioeconomic categories—education, occupational class, and income—provide little evidence of change. There is some suggestion that the higher-status groups have become less successful in controlling fertility and that the various socioeconomic groups are in general becoming more similar in their fertility-planning experience, but the changes are, by and large, weak and irregular. As we have noted elsewhere, comparisons of these categories over 2 Similar results were reported in an analysis of the consistency of the fertilityplanning status classification based on the first and second pregnancies only, and measured across a time interval of three years. See C. F. Westoff, R. G. Potter, Jr., and P. C. Sagi, The Third Child (Princeton: Princeton University Press, 1963), pp. 271-72.

382

CHAPTER IX

time are complicated not only by differences in the level and time pattern of fertility and by the problems of measurement discussed above, but also by selective changes in the composition of the population as well. For example, a higher proportion of women had attended college in 1965 (20 percent) than in 1960 (18 percent) and 1955 (15 percent); Table IX-8. Percent of white couples 18-39 with completely-planned and with excess fertility, by wife's religion and education, by husband's occupation and income, and by the couple's place of residence.

Characteristic

Percent Completely Planned Fertility 1955 1960 1965

Percent Excess Fertility 1955 1960 1965

Number of Couples 1955 1960 1965

Total

19

21

20

13

17

18

2713

2414

2919

Wife's Religion Protestant Catholic

22 10

23 12

23 14

14 11

19 15

19 15

1817 787

1596 668

1907 846

Wife's Education College High School 4 High School 1-3 Grade School

33 23 11 5

35 24 13 6

30 23 12 9

7 10 15 24

11 14 21 32

13 16 23 24

417 1236 681 377

427 1153 579 255

585 1423 643 267

Husband's Occupation Upper white collar Lower white collar Upper blue collar Lower blue collar Farm

29 24 15 12 13

29 24 16 14 24

26 23 18 15 16

10 9 13 18 10

13 17 18 20 21

15 12 21 21 24

620 286 644 765 ,242

725 312 465 670 154

806 391 709 767 130

Husband's Income $10,000+ $ 7,000-9,999 $ 6,000-6,999 $ 5,000-5,999 $ 4,000-4,999 $ 3,000-3,999 $ 0-3,000

22 27 24 18 18 18 16

34 27 17 17 17 20 20

22 20 21 20 18 16 24

7 12 13 12 12 13 16

15 16 22 15 16 16 21

18 19 17 17 15 20 20

88 156 186 393 583 619 581

261 405 312 423 380 306 327

541 731 420 488 302 186 168

Region Northeast North Central South West

20 15 23 17

19 22 25 16

20 19 21 23

10 11 16 15

14 17 18 21

18 17 19 17

669 853 753 433

581 715 744 374

670 896 854 499

Size of Place Cities of 50,000+ and their suburbs Cities under 50,000 Rural nonfarm Rural farm

20 22 14 11

16 22 22 23

22 19 17 18

12 10 20 13

16 23 17 19

18 17 15 25

1569 429 431 284

1594 282 335 203

1826 438 441 214

FERTILITY PLANNING

similarly, the proportion of respondents with husbands in a farm occupa­ tion has progressively declined. The same problem applies to the com­ position of the population by region and by city-size categories, and it is further complicated by selective migration. Analysis of the data by in­ come is especially difficult: average real income has moved up appreci­ ably but there has also been considerable inflation; accordingly, the nominally identical dollar categories are not really comparable. We pre­ sent the tabulation here primarily to provide the data for 1965. One way to evaluate these data is to ask whether the particular inde­ pendent variable, for example education, is more or less predictive in 1965 than in 1960 or 1955 of variations in the dependent variable, such as the proportion of couples with excess fertility. There appears to be a small but consistent tendency for the difference between the proportions of the sample with completely-planned and excess fertility to be a little smaller in 1965 than in 1955 for each variable. For this reason, as well as because of the decline in the proportions of the general population in classes which show deviant proportions in the planning categories (grade school education, farm, low income) there has been movement in the direction of homogeneity of family planning behavior. The trend in fertility planning among nonwhite couples suggests a small increase in the proportion planning fertility completely (Table IX-9) both overall and within each parity. There is also a decrease in the proportion at each parity reporting excess fertility, but the overall pro­ portion shows an increase because of a higher parity distribution in the 1965 than in the 1960 sample. Interpretation of the 1960-65 changes in terms of the other variables in Table IX-9 is complicated by this circum­ stance as well as by the general problems, outlined above, of interpret­ ing these proportions as indices of success in controlling fertility. Fur­ thermore, the 1960 data for nonwhites are based on far too small a sam­ ple to permit more than the broadest generalizations. With all these reservations, we note that the sharp inverse relationship of education and income with excess fertility continues, and that the greatest evidence of excess fertility is still found in the South, especially among couples liv­ ing on farms in this region. SEQUENTIAL PROBABILITIES OF SUCCESS IN CONTROLLING FERTILITY

The preceding sections of Chapter IX have been concerned with the planning statuses of couples classified on the basis of their total repro­ ductive experience to date. The present section is intended to provide

CHAPTER IX

detail concerning the sequence of planning statuses for the separate preg­ nancy intervals which make up that experience. For each closed interval, the planning status was established in the following way. Respondents were identified as users or nonusers of con­ traception during the interval. If nonusers, the outcome of the interval was classified as unexposed (to risk of failure) if the respondent re­ ported that wanting to become pregnant as soon as possible was the rea­ son for not using, as a delay-failure if the respondent wanted to become pregnant some time but not as soon as possible, and as a terminationfailure if she and/or her husband did not want another child. If users, the outcome of the interval was classified as delay-success if the respondent Table IX-9. Percent of nonwhite couples 18-39 with completely planned and with excess fertility, by parity, residence, wife's education, and husband's income.

Characteristic

Percent Completely Planned Fertility 1965 1960

Percent Excess Fertility 1960 1965

Number of Couples 1960 1965

Total

7

11

31.

33

270

837

Parity 2 3 4 5+

5 5 3 0

9 7 5 1

21 39 60 77

19 25 48 69

58 36 30 52

158 158 86 .210

7 12

24 23

4

12 12 22 9

38

18 32 19 39

41 74 19 136

158 131 46 502

O 8

4 11

48 29

50 31

33 237

67 770

Wife's Education College High School 4 High School 1-3 Grade School

24 7 5 1

22 16 3 8

19. 16 38 43

16 28 39 44

37 73 86 74

106 285 289 156

Husband's Income $6,000+ $5,000-5,999 $4,000-4,999 $3,000-3,999 $ 0-3,000

16 6 10 7 4

15 12 8 10 7

16 37 25 27 37

24 27 33 38 46

25 32 51 45 117

183 162 172 156 155

Region Northeast North Central West South Southern Farm Residence On southern farm now All other

Base too small.

*

*

FERTILITY PLANNING

reported stopping contraception in order to conceive, as delay-failure if she reported failure in contraception but signified that the pregnancy was wanted some time, and as termination-failure if she reported failure in contraception and she and/or her husband had not wanted the preg­ nancy some time. For each open interval the planning status was established in the fol­ lowing way. Respondents were classified as users or nonusers of contra­ ception since the termination of the last pregnancy or, if none, since the beginning of that marriage. Users were classified as delayers if they in­ tended to have another child, or as terminators if they intended no more children; nonusers were classified as unexposed if they were not using because they wanted the next child as soon as possible, as delayers if they intended to have another, but not as soon as possible, and as terminators if they intended no more children. In summary, the classification subdivided respondents in each interval according to intention (unexposed, delayer, terminator), according to contraceptive practice (use or nonuse), and according to outcome (suc­ cess, failure, open). Because of insufficient sample sizes at higher orders, the presentation is limited to data for the first five pregnancy intervals. The order of each interval is the order of the pregnancy that closed it or, if the interval is open, would have closed it. Although the analysis takes the data at face value, as if they were com­ pletely accurate, various kinds of evidence make it apparent that every element of the classification scheme contains errors of reliability and validity. Some pregnancies which ended in fetal death have been omitted from respondents' reproductive sequences. Some contraceptive use as well as nonuse in particular intervals has been concealed or forgotten. Although we treat the intention as if it were prior to contraceptive prac­ tice and outcome, the information is all retrospective; it appears likely that the outcome has sometimes influenced the respondent to modify her report of intention and contraceptive practice. Appendix A contains sug­ gestive information on the reliability of variables used here, but that does not exhaust the sources of difficulty because the reinterview procedure cannot reveal consistent misstatement. In short, the evidence presented here (as elsewhere in this study) should be treated with considerable circumspection. The data to be presented are defined somewhat differently from those in the rest of this book. In the first place, to enlarge the available subsample sizes, the scope of the inquiry has embraced all women born since July 1, 1910 (rather than, as is customary for most of the book,

CHAPTER IX

those born since July 1, 1920). Secondly, intervals are excluded if they encompass a marital dissolution and remarriage, if they are terminated by the sterilization of either spouse, or if the reason for nonuse (in the open interval) is sterility. Finally, in order to avoid overrepresentation of respondents with short interval lengths, intervals which began less than 4 years prior to mid-1965 were excluded. This interval of a little less than 4.5 years prior to interview was an arbitrary compromise be­ tween reduction of the indicated bias and the concomitant sacrifice of data. A further note of explanation is appropriate concerning the N's which are provided for every table in this section. As discussed in Chapter I, our sample was double-weighted for Negroes and half-weighted for women aged 45-54. The computer program used for the present section restored the weighting appropriate to the universe. The N's reported be­ low are the weighted N's; they are, on the average, approximately twice as large as those unweighted N's which would be appropriate for sig­ nificance tests. As a lower limit for reporting proportions in the tables, we have used an N of 45 for whites and 25 for Negroes. One further dif­ ference from the rest of our tables is that the totals reported here exclude that small part (approximately 1 percent) of the sample which was nonNegro and nonwhite. Planning Statuses of Individual Intervals

To give a sense of the kinds of information available for the present discussion, we have prepared Tables IX-10/IX-15, summarizing the dis­ tribution of the sample by planning status in the first five intervals. The classification can be considered as consisting of three parts in causal order (assuming reliable reporting). First is the reproductive intention: to delay, to terminate, or to remain unexposed (to risk of failure) by not using contraception because a baby is wanted as soon as possible. Second is the means used to achieve that end: either use or nonuse of contracep­ tion. Finally there is the outcome of the interval: success, failure, or indeterminate (i.e., an open interval) at interview. Note that those who are unexposed must be nonusers and cannot be failures; those who are terminators cannot be successes; and those who do not use contracep­ tion despite the intention to delay or terminate cannot be successes. The proportions intending to terminate in each interval order are shown in Table IX-10, by race, religion, and education of wife. The pro­ portion for the total sample is low in the first two intervals, and then rises progressively to 50 percent in interval five. In every interval order, the

FERTILITY PLANNING Table XX-10.

Interval X

II

III

IV

V

Percent intending to terminate, by race, religion, and education.

Race and Religion

Percent Total College HV S. 4 High School 4 High School 4 < High School 4

> High School 4 High School 4 < High School 4

> High School 4 High School 4 < High School 4

> High School 4 High School 4 < High School 4

17 or Less

18 - 20

21 - 23

24 or More

45 (219) 38 (267) 34 (272)

55 (513) 47 (1209) 50 (1926)

60 C680) 63 High School 4 High School 4 < High School 4

Total

II

I

Education

48 (133) 31 (204) 34 (283)

29 (121) 24 (186) 34 (242)

— (43) 32 (114) 37 (174)

66 (58) 45 (175) 36 (261)

53 (86) 48 (307) 41 (389) 61 (78) 35 (215) 34 (196)

— (26) 63 (133) 51 (600)

48 (248) 41 (608) 43 (1277)

IV

(30) 62 (184) 56 (730)

54 (327) 44 (910) 46 (1598)

III

14 (55) 17 (119) 38 (155)

— (17) 26 (97) 21 (122)

(35) 25 (74) 40 (215)

— (37) 51 (99) 54 (433)

36 (144) 30 (389) 44 (925)

V

Percent pregnant within the first year of nonuse intervals, by education and initial age

(+2(Ν -1)).

Initial Age

Table XI-44.

BIRTH AND PREGNANCY INTERVALS

make runs for race and religion by education (tabulations not shown here) and noted two interesting findings. In every order the level of fecundity was (1) higher for college-educated Catholics than for collegeeducated non-Catholics, and (2) higher for college-educated whites than for college-educated Negroes. The differences by race and religion were negligible in the group with less than 4 years of high school. COMPARATIVE EFFICACY BY RACE, RELIGION, AND EDUCATION

In studying differences in fertility by variables like race, religion, and education, there are three questions which, in sequence, need to be answered, in order to determine the sources of these differences. The first is the intention of the respondents—do they want to have a child as soon as possible, do they want a child but only after a delay, or do they not want another child at any time? The next is, given that the achievement of their goal requires use of contraception, to what extent do they use contraception? Finally, we need to know, if they choose to use contra­ ception to achieve their reproductive goals, how successful are they in that endeavor? This section assembles evidence relevant to the answer­ ing of these three questions. In Table XI-45 we show the distribution of the various subcategories of race, religion, and education, by the three categories of intention, within each of the first five pregnancy orders. The proportion wanting the next pregnancy as soon as possible declines steadily from the first to the fifth order and is systematically higher for Catholics than for nonCatholics. Of more interest is the proportion wishing to terminate with a particular pregnancy. In every order this proportion is higher for Negroes than for whites, and higher for white non-Catholics than for white Catholics. Both observations hold at all levels of education. Thus, in terms of desired family size, we find a distinct stratification: white Catholic high, white non-Catholic in the middle, and Negro low. There is little patterning of termination intentions by education except that, for the Catholics, the lowest education group has the highest termination in­ tention (the lowest desired number of pregnancies), whereas, for the Negroes, the middle education group has the highest termination intention. In the next table (Table XI-46), we examine the extent to which those who do not register their intention as nondelay, that is, those who need to use contraception in order to achieve their reproductive objective, in fact use contraception. The proportion ranges between 80 and 85 per­ cent, with little difference from order to order. The whites are much

T WNC WC W N

V

Weighted H's.

ft

16 13 21 16 15

T WNC WC W N

IV

14 13 17 14 13

17 13 23 16 18

III T WNC WC W N

33 31 40 34 28

42 41 47 43 35

53 54 55 54 . 44

66 69 63 67 53

22 19 29 22 24

T WNC WC W N

II

59 64 49 60 50

53 56 44 52 58

42 46 32 41 50

30 33 22 30 38

12 12 8 11 23

7 7 6 7 12

Total DEL TER

3945 2226 1112 3338 547

11 8 19 11 0

16 13 24 16 10

14 12 23 15 12

8624 5409 2166 7575 929

6095 3631 1650 5281 730

21 17 38 21 19

25 20 43 24 31

37 33 44 37 47

45 44 47 45 43

57 58 58 58 50

69 73 54 70 58

71 73 53 71 60

1406 1040 272 1312 84

28 30 19 28 38

52 58 37 52 53

39 43 29 39 47

11 9 16 11 6 10 7 13 10 7

929 658 214 872 55 581 388 156 544 35

13 11 20 14 9

17 14 25 17 17

1719 1264 320 1584 115

10 9 7 9 23

5 7 4 5 9

32 26 45 31 32

ND

1941 1452 358 1810 109

Hish School 4 or More DEL ND TER Nft

10010 6322 2450 8772 1097

9765 6490 2428 8918 721

Na

36 36 41 38 23

46 44 53 47 30

56 56 60 57 51

72 74 69 73 59

62 68 49 62 56

54 57 46 53 70

43 47 32 41 64

30 34 21 29 40

11 12 6 10 23

6 6 6 6 12

High School 4 DEL TER

1412 802 496 1298 100

2362 1419 764 2183 155

3587 2309 1034 3343 214

4219 2738 1180 3918 269

4303 2894 1176 4070 197

NA

18 18 19 19 16

20 17 28 20 18

21 17 27 20 21

28 26 32 27 27

41 39 47 41 40

30 28 37 31 28

37 38 39 38 35

48 50 47 49 41

57 60 58 60 50

50 52 47 51 46

52 54 44 51 56

43 46 34 42 47

31 33 25 30 37

15 14 10 13 24

9 9 6 8 13

1952 1036 460 1496 412

2804 1554 672 2226 520

3631 2060 860 2920 631

4069 2320 950 3270 713

3521 2144 894 3038 415

Less Than Hiah School 4 ND DEL TER N

Percent distribution of intention -- nondelay (ND), delay (DEL), terminate (TER) -- by race, religion, and education.

34 29 45 33 37

T WNC WC W N

I

ND

Table XI-45.

BIRTH AND PREGNANCY INTERVALS Table XI-46,

Race and Religion I

II

III

IV

V

Percent of those intending to delay or terminate using contraception, by race, religion, and education.

Total

> H.S. 4

H.S. 4

< H.S.

Total

81

89

84

70

WNG WC

84 75

91 80

86 79

75 69

W N

82 64

89 86

84 82

73 48

Total

84

91

89

75

WNC WC

88 81

94 79

91 86

81 75

W N

86 66

92 79

90 84

79 57

Total

85

93

91

75

WNC WC

87 86

96 89

92 89

78 79

W N

87 68

94 80

91 85

78 59

Total

83

92

90

73

WNC WC

86 83

94 90

92 86

77 76

W N

85 68

93 83

90 85

77 61

Total

80

91

90

68

WNC WC

84 81

94 86

92 86

73 72

W N

83 64

92 84

89 91

73 54

more likely to use contraception than the Negroes (the absolute differ­ ences over the various orders average 19 percent), and the non-Catholics somewhat more likely to use than the Catholics (an average absolute dif­ ference of 5 percent). There is a small difference between those who en­ tered college and those who completed high school, but a large difference between the latter and those who did not complete high school. These educational differentials hold within racial and religious subgroups. The case of the Negroes is interesting: those who completed high school are a little more likely to use contraception than those who entered college, and much (a mean absolute difference of 25 percent) more likely to use

CHAPTER XI

than those who did not complete high school. When we compare whites with Negroes, with education controlled, we find an especially large dif­ ference at the lowest educational level; when we compare white Catholics with white non-Catholics, the largest difference is at the college level. Over the five orders examined, these results are strongly patterned; they give confidence in the reliability of our output. Whatever the level of use, it is to no avail if the use is ineffectual. Be­ cause we have become sensitized to the distinction between efficacy when the purpose is to delay and efficacy when the purpose is to terminate, we have prepared Tables XI-47 and XI-48 showing, respectively, the perTable XI-47. Percent of users intending to delay who fail within the first 2 years", by race, religion, and education.

Race and Religion I

II

V

46 (4667)

> H.S. 4

H.S. 4

< H.S. 4

41 (2210) 38 (1652) 46 (456) 40 (2108) 65 (92)

62 (1260)

Total WNC WC

43 (3480) 52 (890)

W N

45 (4370) 65 (251)

40.(1197) 36 (974) 57 , (150) 39 (1124) 53 (57)

Total

42 (5477)

35 (1067)

WNC WC W N

38 (3786) 44 (1242) 40 (5028) 68 (391)

30 (860) 60 (138) 34 (998) 43 (53)

37 (2678) 33 (1820) 42 (704) 36 (2524) 64 (136)

W N

39 (3579) 35 (2307) 41 (970) 37 (3277) 65 (260)

34 27 52 33 52

(696) (520) (136) (656) (32)

37 (1661) 32 (1043) 44 (520) 36 (1563) 51 (88)

Total WNC WC

37 (1923) 35 (1145) 34 (600)

41 (353) 35 (240) 55 (92)

35 (869) 34 (505) 32 (320)

W H

35 (1745) 66 (160)

41 (332) (19) -

34 (825) 69 (38)

Total

44 (923) 39 (508) 44 (326)

38 (165)

WNC WC W H

41 (834) 71 (85)

41 (395) 28 (226) 56 (152) 40 (378) (17)

III Total WNC WC

IV

Total

Weighted N1s in parentheses.

33 (98) 40 (56) 36 (154) (H)

61 (854) 59 (284) 61 (1138) 69 (102) 53 (1732) 52 (1106) 42 (400) 49 (1506) 77 (202) 44 (1222) 42 (744) 32 (314) 40 (1058) 76 (140) 38 35 27 33 68

(701) (400) (188) (588) (103)

50 (363) 55 (184) 31 (118) 45 (302) 75 (57)

BIRTH AND PREGNANCY INTERVALS

cent of users intending to delay and the percent intending to terminate who fail within the first 2 years. Table XI-47 indicates the efficacy of use to delay. The proportion failing within 2 years varies little by order, around an average of 42 percent. The racial differences in efficacy are very pronounced, ranging between 20 and 31 percent in absolute terms. Similarly, the white Catholics have higher failure rates than the white non-Catholics, although not in such pronounced degree (an average absolute difference of 5 percent). The difference between those with at least some college education and those who have completed high school is small (but positive); the difference between the latter and those who have less than 4 years of high school is large (although it declines with advancing order). For all levels of education, the racial difference holds, and it is particularly great among those with less than 4 years of high school. An interesting finding turns up by religion when education is con­ trolled: the differences between college-educated Catholics and collegeeducated non-Catholics is large and positive; the difference between Catholics and non-Catholics at the grade-school level is appreciable and negative. The general tendency for efficacy to vary directly with educa­ tion holds for whites and Negroes considered separately, and for white non-Catholics. The reverse is true for white Catholics: broadly speaking, those with the least education show the highest efficacy and those with the most show the lowest efficacy. This kind of result has appeared re­ peatedly in this study; it is apparently a reflection of the circumstance that college-educated Catholics are the most likely to adhere to the Church precepts concerning contraception, that is, most likely to use the less efficacious rhythm method. More important, because more consequential for a couple's future, is the level of efficacy when the intention of contraceptive use is to termi­ nate childbearing. The data, assembled in Table XI-48, are limited to the third, fourth, and fifth orders because of insufficient cases in the lower two orders. By race there are very large differences: in absolute terms, 40 percent in the third order, 33 percent in the fourth order, and 22 per­ cent in the fifth order. These differences persist within educational cate­ gories, insofar as the data permit us to examine the point. By religion the differences are negligible, overall and within educational categories. There is a small difference, essentially confined to the fifth order, be­ tween the level of efficacy for those who have at least entered college and those who have completed college; there is a large difference between those who did and those who did not complete high school. In summary, the proportion intending to terminate is higher for

CHAPTER Xl Table XI-48. Percent of users intending to terminate who fail within the first 2 years , by race, religion, and education.

Race and Religion III Total

XV

V

Total

> H.S. 4

H.S. 4

< H.S. 4

32 (1972)

25 (341)

26 (880)

41 (751)

WNC WC

28 (1382) 26 (356)

20 (276) (40)

23 (636) 21 (168)

38 (470) 28 (148)

W N

27 (1738) 67 (216)

23 (316) (23)

22 (804) 68 (66)

36 (618) 70 (127)

Total

33 (1855)

23 (305)

25 (805)

46 (745)

WNC WC

27 (1224) 33 (370)

20 (236) 25 (50)

19 (524) 29 (190)

41 (464) 42 (130)

W N

29 (1594) 62 (23?)

21 (286) (19)

22 (714) 55 (81)

41 (594) 68 (139)

Total

33 (1416)

15 (236)

30 (584)

42 (596)

WNC WC

29 (868) 31 (338)

7 (176) (42)

29 (346) 26 (174)

40 (346) 32 (122)

W N

30 (1206) 52 (200)

15 (218) (16)

28 (520) 49 (62)

38 (468) 57 (122)

Weighted N's in parentheses.

Negroes than for whites, but the proportion using and the proportion succeeding with use—whether for delay or for termination—are much higher for whites than for Negroes. Thus Negroes end up with higher fertility than whites because, although they desire fewer children, they are less likely to use contraception and less successful when they do use. White non-Catholics are less likely than Catholics to want the next preg­ nancy as soon as possible, and more likely to want to terminate at each pregnancy. Moreover, white non-Catholics are more likely to use, and, if using, more likely to be successful. The relationships by education are more complex. Except for Catholics, the differences in use and efficacy are small between those with college educations and those who have graduated from high school, but large between either of these and the respondents who did not complete high school. Because of concern about the extent to which initial age may be an important hidden influence on the relationship between efficacy, on the one hand, and race, religion, and education, on the other (in the light of the large differences in the initial-age distribution shown in Table XI-41 above), we have prepared Table XI-49, to show failure rates for the

43

40 70

36 38 53

49

45 55

47 66

40 45 64

Total

White Non-Catholic White Catholic

White Negro

> High School 4 High School 4 < High School.4

32 56 22 32 43

30 64 30 29 41

32 65 41 36 42

36 64

33 45

39

194 539 344

481 630 204

444 407 1146 1187 419 482

366 696 257

319 640 279

841 340

106 277 214 299 894 424

1275 1709 1069 689 439 2390 3033 2673 1760 1033 1396 2143 2015 1533 1028

894 344

1240 1238 1181 55 71 77

938 302 524 69

> High School 4 High School 4 < High School 4

33 26 31

28 57

26 31

960 101

1878 1912 1470 111 147 128

4730 5790 5215 3537 2190 281 531 481 401 292

White Negro

36 31 38

33 43

28 46

29

478 35 126 233 160

202 427 202

330 148

531 254 785 42

519

25 29 38

31 40

30 32

31

V

831

30 21 33

24 60

23 24

26

21 - 23 III IV

1315 1319 1238

40 43 60

43 62

39 54

34

II

348 176

32 34 46

36 53

33 42

44

I

692 268

1526 1464 1084 352 448 386

3744 4406 3863 2533 1474 986 1384 1352 1004 716

White Non-Catholic White Catholic

41 32 38

32 65

29 40

38

V

597

Weighted N'S

42 42 44

41 67

38 50

43

36

18 - 20 IV III

Percent

II

2009 2076 1617 1077

46 47 61

49 71

46 59

50

I

5061 6348 5757 3982 2500

30 31 42

31 34

29 33

31 36

V

35

IV

34

35

Total III

Total

38 45

II

I.

Race, Religion, and Education

Table Xl-49. Percent of users falling within the first 2 years, for the total sample and for initial ages 18-20 (+2(N-I)) and 21-23 (42(N—1)), by race, religion,-and education.

CHAPTER XI

three variables when initial age is controlled. What we find is that the previously observed relationships by religion become larger—because white Catholics, who have the lower efficacy, have the higher initial-age distribution, and initial age is directly related to efficacy—and that the previously observed relationships by race become smaller—because the Negroes, who have the lower efficacy, have the lower initial age distribu­ tion. Education is the variable most closely linked to initial age; the (admittedly small) advantage in efficacy of women entering college over those who have only completed high school is washed out when initial age is controlled. The group which did not complete high school remains, however, the educational group with the highest failure rates. SUMMARY

In this chapter we present two approaches to the study of the timing of fertility for cohorts and subgroups with differentially incomplete re­ productive experience. The first approach, examination of birth intervals, starts with the beginning of the reproductive sequence (or at least the conventional beginning for most women)—the age at first marriage. A trend toward earlier marriage has been accompanied by, but only partly explained by, a trend toward a larger proportion of marriages reported to be preceded by pregnancy. White Catholics marry later than white non-Catholics, and whites marry later than Negroes. Age at marriage varies strongly and directly with education, and the causal relationship probably works both ways. Of all first births, one-ninth for the whites but one-half for the Negroes are the outcome of premarital pregnancy. We have examined birth intervals up to the fourth birth. Over the time span represented in this study, the first three intervals have become dis­ tinctly shorter. Intervals tend to be shorter for Catholics than for nonCatholics (within the white sample), shorter for Negroes than for whites, and shorter for those with less education. The problem of terminat­ ing childbearing is aggravated for Negroes by the fact that they are pro­ gressively younger than whites when they reach each higher parity. Since the precautions required to guard against biased birth-interval measurement proved to be wasteful of data, we proceeded with our sec­ ond approach to the subject: the study of pregnancy intervals by lifetable procedures. We present efficacy measures for use intervals, and fecundity measures for nonuse intervals, for the first five pregnancy inter­ vals, and for the successive segments in each interval, in three forms: the central pregnancy rate, the proportion surviving to each exact "age," and the mean interval-length for those who became pregnant by each exact

BIRTH AND PREGNANCY INTERVALS

age. In every order, less than one-half of the users are successful in avoiding pregnancy for at least 5 years. The proportions "surviving" in nonuse intervals are much lower, but high enough to provoke the suspi­ cion that there is substantial concealed use. Because of the dependence of the decision to use contraception on the perception of fecundity, the measurement of fecundity and of efficacy are conceptually interrelated. Those using contraception to delay are somewhat less successful than those using to terminate before the third, fourth, and fifth pregnancies. The successful delayers show very high fecundity upon stopping contra­ ception in order to conceive, so high indeed that we suspect the accuracy of some of the reports. We report three trends: more use of contraception; less effective use; and higher fecundity among nonusers. The measurement of trends from data for a single cross-section of respondents poses many difficulties. Those attributable to the truncated nature of such a sample—the bias to­ ward lower initial ages for each interval—we have been able to control, but about those attributable to misreporting we can only speculate. We also show that the pill is by far the most efficacious method in common use, and rhythm clearly the least, with diaphragm, condom, and with­ drawal all only moderately effective. Data on various factors which may affect fecundity are presented. We have determined that nonusers who began their intervals by breastfeed­ ing had a distinctly later time-pattern of pregnancy rates than did nonlactators, and that the delay in resumption of full fecundity appeared to be directly related to the duration of lactation. We have observed an in­ verse relationship between fecundity and age at menarche (perhaps more than coincidentally linked with a previous finding of a weak direct rela­ tionship between age at marriage and age at menarche). A comparison of the experience, in both nonuse and use intervals, of the respondents who reported douching, but not for purposes of regulating fertility, reveals lower pregnancy rates for them than for the rest of the sample. Despite the inconsistency in reporting coital frequency, and despite the time lapse between the intervals examined and the month preceding the interview (the referent for the question), we have found a (weak) posi­ tive relationship between coital frequency and the pregnancy rate in nonuse intervals. Finally, patterns of intention, contraceptive use, and efficacy have been examined for whites and Negroes, and for white Catholics and white non-Catholics, at three educational levels. The much higher fertil­ ity of Negroes than of whites can be explained not by desire—since Ne-

CHAPTER Xl

groes intend to terminate their childbearing in a lower order than whites —but by a much lower disposition to use contraception to achieve that objective, and a much lower efficacy when contraception is used. The much higher fertility of white Catholics than of white non-Catholics, on the contrary, is explained not only by somewhat less use and somewhat less efficacy in use but also by their intention to terminate at a higher parity than white non-Catholics. The data reported in this chapter are probably the most crucial to our understanding of the processes underlying fertility but at the same time the most fraught with possibilities of misinterpretation because of unre­ liability. The subtleties of complex explanatory systems, employing sophisticated social, economic, and psychological concepts, will provide meagre returns unless and until we learn how to measure more ac­ curately the dependent variables at the core of the analysis of fertility.

APPENDIX A

Inconsistency of Response BECAUSE of concern with the reliability of the data collected in fertility studies and with the analytic implications of unreliability, and because there is so little pertinent information in the literature,1 we designed a study to reinterview a probability subsample of our respondents some three months after the initial interview. A total of 440 women were suc­ cessfully reinterviewed, some 81 percent of the 540 sought.2 The reinter­ view was conducted in the same manner as the original interview, with only an additional explanation (if necessary) that sometimes more infor­ mation is needed in such surveys or a vague statement that the reinter­ view involved methodological aspects of survey research. Interviewers were not selected in any special way; there was no attempt made to ar­ range for an interviewer to recontact the same women she had inter­ viewed earlier, and, in fact, most respondents had a different interviewer. The reinterview included some 60 percent of the questions asked in the original interview, concentrating on those items of most significance for fertility studies. This appendix consists of an analysis of data for the 382 women under age 45 who were reinterviewed. Information gathered in this way does not precisely concern reliability, although it is obviously highly pertinent. What we have literally been able to observe is merely whether the information punched from the sec­ ond interview is the same as or different from that punched from the first interview. Thus this appendix is entitled "Inconsistency of Response," with due recognition that inconsistency may signify a real change during the several months rather than the unreliability of either response. Of course, consistency cannot be taken as a guarantee of the value of re­ sponses—since stereotypical responses would be registered as consistent, however meaningless, and intentional misstatements could be repeated. Furthermore, we have not explored the sources of inconsistency, which could occur not only as a consequence of the respondent's words or in­ terviewer's bias but also through errors in recording, editing, coding, or punching. It is also probable that the instrument, although we retained the original wording of each individual question, becomes somewhat 1 But see Charles F. Westoif, Robert G. Potter, and Philip C. Sagi, "Some Estimates of the Reliability of Survey Data on Family Planning," Population Studies 15 (July, 1961): 52-69. 2 The largest category of nonrespondents (42) was "refusal." The remainder were "not at home" (19), "moved" (11), and "all other reasons" (28).

APPENDIX A

transformed by the abbreviation of the total length. Moreover, our esti­ mates of inconsistency might have been somewhat different had they been based on reinterviews after shorter or longer intervals than was actually the case. Despite these limitations on the present section as a methodological contribution, we are interested in the extent of incon­ sistency as an elementary clue to the existence of statistical problems which detract from our findings in various ways. MEASUREMENT OF INDIVIDUAL INCONSISTENCY

By inconsistency we mean here simply the difference between the two responses (at two times, not far apart) to the same question by the same respondent. Our concern encompasses both individual and aggregate inconsistency: the present section is devoted to the former. It is no simple matter to devise a satisfactory measure of individual inconsistency. Obviously the proportion giving a different response is not satisfactory, since this proportion is reduced by any tendency for random correspondence. For some variables at refined levels of measurement, an obvious suggestion is the coefficient of correlation, but we did not choose this measure. For one thing, the coefficient of correlation is highly sensitive to skewness in the marginal distributions, and these are com­ mon in our data. Secondly, we wanted a measure which could be applied uniformly to all of our variables, whether or not they happened to be nominal, ordinal, or interval scales. The measure we have developed can be explained and justified in the following way. Let the proportion responding consistently, that is, giving the i-th response on both occasions, be Pi. Let the marginal proportion giving the i-th response on both occasions, but with random relationship be­ tween the first and second response, be ¢¢. Then the observed proportion giving the same response, say O, is %Pi +J!£i!, and the expected proportion giving the same response, Sgi under random conditions, say E, is 2 (Pi +